Ontents: SPECIAL TOPIC: Packet Transport Networks

Contents
www.zte.com.cn/magazine/English ISSN 1673-5188

CODEN ZCTOAK
(英文版)
ZTE COMMUNICATIONS
Http://www.zte.com.cn/magazine/English September 2010, Vol.8 No.3
Email: magazine@zte.com.cn SPECIAL TOPIC: Packet Transport Networks
Editorial Committee Members (in Alphabetical Order):

Chairman: Zhong Yixin Ai Bo, Cao Shumin, Chang Jinyun, Jiang Lintao, Lei Zhenzhou, Xu Chengzhong, Xu Heyuan, Yang
Chen Changjia, Chen Jianping, Li Hongbin, Li Jiandong, Li Lemin, Yixian, Yang Zhen, Yin Yimin,
Vice Chairmen: Hou Weigui, Chen Jie, Chen Xisheng, Li Shaoqian, Li Xing, Meng Luoming, You Xiaohu, Yue Guangxin,
Mi Zhengkun Cheng Shiduan, Cheng Shixin, Mi Zhengkun, Ni Qin, Sun Zhenge, Zhang Tongxu, Zhang Zhijiang,
Gao Wen, Gong Shuangjin, Tan Zhenhui, Tian Wenguo, Zhao Houlin, Zhao Huiling,
Gu Yongcheng, Gu Wanyi, Wang Xiaoming, Wang Xiaoyun, Zhao Xianming, Zhong Yixin,
Guo Yunfei, Hou Weigui, He Shiyou, Wang Yumin, Wei Leping, Wei Guo, Zhou Susu, Zhu Jinkang
Hong Bo, Ji Yuefeng, Jiang Hua, Xie Daxiong, Xie Xiren, Xu Anshi,
SPECIAL TOPIC: Packet Transport Networks

ZTE COMMUNICATIONS
Vol. 8 No.3 (Issue 27)
1 Study of the China PTN Industry Standard and Its Key Issues
Quarterly 5 Standardization Progress of Packet Transport Networks
First Issue Published in 2003
9 Several Issues in the Development of Packet Transport Networks
Supervised by:
Anhui Science and Technology 13 Features of PBB-TE Architecture and GMPLS Control Technology
Department 17 Ring Protection and Survivability Mechanisms for Packet Transport Networks
Sponsored by: 22 Service Adaptation and Label Forwarding Mechanism for MPLS-TP
ZTE Corporation and Anhui Science
and Technology Information 27 PTN Clock Synchronization Technology and Its Applications
Research Institute 32 PTN and IP-Based Mobile Backhaul
Staff Members: 37 Analysis of Time Synchronization in PTN
Editor-in-chief: Xie Daxiong
Deputy Editor-in-chief: Deng Xin
Executive Deputy
RESEARCH PAPERS
Editor-in-chief: Huang Xinming
Editor in Charge: Zhu Li 42 Vehicular Ad-Hoc Networks: An Information-Centric Perspective
Editors: Paul Sleswick, Yang Qinyi, Lu Dan
50 Applying Analytical Hierarchy Process to a WiMAX Performance Evaluation
Producer: Yu Gang
Circulation Executive: Wang Pingping Model
Editorial Correspondence:
Add: 450 Rongshida Avenue,
Hefei 230041, P. R. China
DEVELOPMENT FIELD
Tel: +86-551-5533356 56 Research on Convergence Network of EPON and WiMAX Based on ROF
Fax: +86-551-5850139
Email: magazine@zte.com.cn
Published and Circulated LECTURE SERIES

(Home and Abroad) by: 59 Cloud Computing (3)
Editorial Board of
ZTE COMMUNICATIONS
Printed by: ROUNDUP
Hefei Zhongjian Color Printing Company
Publication Date: September 25, 2010 12 ZTE France Awarded Best Investor Prize 2010
Publication Licenses: 21 ZTE Tops the Global CDMA Market with a 30% Market Share
ISSN 1673-5188
CN 34-1294/TN 41 ZTE Positioned in the Leaders Quadrant in Leading Industry Analyst Firm Report
Advertising License:
49 ZTE and Three Release Android Éclair Handset in the UK
皖合工商广字 0058 号
Annual Subscription Rate:
USD$20
DEPARTMENTS
Responsibility for the contents rests 64 Ad Index
upon authors of signed articles and
65 Abbreviation Index
not upon the editorial board of
ZTE COMMUNICATIONS or its sponsors.
All rights reserved.
D:\EMAG\2010-08-27/VOL8\CONTETN.VFT——1PPS/P 1
Study of the China PTN Industry Standard and Its Key Issues Special Topic
Zhang Haiyi, Li Fang
Study of the China PTN Industry

Standard and Its Key Issues
(Institute of Communication Standards Research, China Academy of Telecommunication Research of MIIT, Beijing 100191, P. R. China)
Abstract:
In the area of carrier technologies, Packet Transport Networking (PTN) continues to attract significant research interest. Although
international standards are not yet complete, China is leading the way in developing PTN equipment and network applications. General
Technical Requirements for Chinese Packet Transport Networks is the PTN industry standard in China defined by China Communications
Standards Association (CCSA). This article introduces the background of PTN technology and standards, then illustrates four important
aspects: PTN network architecture, multi-services bearing and data transfer function, PTN network protection, and OAM architecture and
functional requirements. Taking the profit motives of Chinese carriers and vendors into consideration, some key issues in PTN technology
selection, network applications, and future evolution are discussed.
private lines/networks. The MANs of 802.1Qay—a connection-oriented

1 Implementation of Packet Chinese operators are being Ethernet transport technology. Also
Transport Networks transitioned away from TDM to PTN called Media Access Control (MAC) in
P
acket Transport Networking packet switching. MAC, Provider Backbone Bridge (PBB)
(PTN) is Connection-Oriented PTN has two implementation represents an improvement on IEEE
Packet-Switched (CO-PS) technologies: MPLS Transport Profile 802.1ah, eliminating connectionless
multi-service unified transport (MPLS-TP) and Backbone Bridge Ethernet characteristics such as MAC
technology. PTN is sufficient for bearing Traffic Engineering (PBB-TE). Based address learning, spanning tree
carrier-class Ethernet services, and on Internet Protocol/Multiprotocol Label protocol, and flooding. Traffic
features standardized services, high Switching (IP/MPLS), MPLS-TP Engineering (TE) is also incorporated to
reliability, flexible scalability, strict discards hop-by-hop forwarding enhance QoS. PBB-TE currently
Quality of Service (QoS), and improved according to IP addresses and supports Point-to-Point (P2P) and
Operation, Administration and establishes transport paths that are Point-to-Multipoint (P2MP) connection
Maintenance (OAM). Moreover, it independent of the control plane. The service transport and linear protection.
supports legacy Time Division connection-oriented end-to-end label However, it does not yet support
Multiplexing (TDM) and Asynchronous forwarding capability of MPLS is multipoint-to-multipoint
Transfer Mode (ATM) services, and retained, and non connection-oriented connection-oriented service transport
inherits management functions such as and non end-to-end and ring protection.
graphical interface and end-to-end features-including Penultimate Hop MPLS-TP and PBB-TE differ in terms
configuration of Synchronous Digital Popping (PHP), Label Switched Path of data forwarding, multi-service
Hierarchy (SDH) network management. (LSP) Merge, and Equal-Cost bearing, network protection, and OAM
PTN is currently applied in Metropolitan Multi-Path Routing (ECMP)—are mechanism[1-2]. As far as
Area Networks (MANs) that bear eliminated. MPLS-TP therefore standardization, vendor products,
services with QoS requirements—such supports definite end-to-end transport operator applications, and the broader
as mobile backhaul and enterprise paths, PTN-style network protection, industrial chain are concerned,
and OAM capability. MPLS-TP seems to have a more
This work was funded by the National High The other implementation technology promising future than PBB-TE.
Technology Research and Development
Program of China “863”
( program) under Grant is Backbone Bridge Traffic Engineering Therefore, MPLS-TP is currently
No. 2007AA01Z253. (PBB-TE) specified by IEEE treated as the leading PTN (4) PTN
September 2010 Vol.8 No.3 ZTE COMMUNICATIONS 01
Special Topic Study of the China PTN Industry Standard and Its Key Issues
following functional requirements for

Customer/Client Service Layer bearing PW-based emulation services:
(1) Pseudo Wire Emulation
Channel Virtual Channel Edge-to-Edge (PWE3) encapsulation
should be used uniformly to bear
Path Virtual Path (Optional) emulation services. The PWE3 control
word should follow RFC4385.
Virtual Section (Optional) PTN (2) TDM service emulation and
Hierarchy transport should be supported. Specific
GFP requirements are as follows:
Transmission Circuit Switching Technology •TDM service bearing in the
Media (OTN, SDH, PDH)
Physical Media (802.3) Structure-Agnostic TDM over Packet
Physical Media (SAToP) mode should be supported,
(OTM-n, STM-N , DSn /En, xDSL) with the PWE3 encapsulation and
GFP: Generic Framing Procedure PDH: Plesiochronous Digital Hierarchy STM: Synchronous Transport Module control word following RFC4553. The
OTM: Optical Transport Module PTN: Packet Transport Network SAToP mode can be applied to any
OTN: Optical Transport Network SDH: Synchronous Digital Hierarchy
signal structure of TDM circuit
▲Figure 1. PTN layers. emulation.
•Support for TDM service bearing in
implementation technology. processing. The protection mechanism Circuit Emulation Services over Packet
does not rely on IP packets; Switch Network (CESoPSN) mode is
2 Main Content of the •Support for bidirectional and optional, with the PWE3 encapsulation
unidirectional point-to-point transport and control word following RFC5086.
China PTN Industry paths, as well as TE control of P2P and CESoPSN is mainly used for Nx64kb/s
Standard P2MP transport paths. signal structures of TDM circuit
General Technical Requirements for As shown in Figure 1[3], PTN has three emulation, and is capable of saving
Chinese Packet Transport Networks layers: Virtual Channel (VC) layer, bandwidth.
specifies PTN architecture, Virtual Path (VP) layer, and Virtual •Support for TDM service bearing in
multi-service bearing and data Section (VS) layer. At the bottom of PTN the SDH Circuit Emulation over Packet
forwarding functions, network is the physical media layer, (CEP) mode is also optional, with PWE3
protection function and performance, implemented by IEEE802.3 Ethernet/ encapsulation and control word
OAM capability, QoS, packet SDH and Optical Transport Network following RFC4842. CEP is mainly used
synchronization, network interface, (OTN) Connetion-Oriented for VC12/VC-4 signal structures of
network performance, network Circuit-Switched (CO-CS) TDM circuit emulation.
management, and functions of the technologies. In a MPLS-TP system, a (3) Ethernet L2 service emulation and
control plane. The following introduces Pseudo Wire (PW) layer is equivalent to transport should be supported. The
main features of PTN. PTN VC, and the LSP layer is equivalent specific requirements are as follows:
to PTN VP. •Ethernet Line (E-Line), Ethernet
2.1 PTN Architecture Local Area Network (E-LAN), and
PTN has the following technical 2.2 PTN Multi-Service Bearing and Ethernet Tree (E-Tree) services are
characteristics: Data Forwarding Functions supported, and should follow the
•CO-PS, packet switched kernel, The client layer in MPLS-TP related specifications of International
and multi-service bearing; architecture[4] includes PW-based Telecommunication Union’ s
•It is strictly connection-oriented, emulation services, services with MPLS Telecommunication Standardization
and has long-lasting connection that labels, and IP services. The standards Sector (ITU-T) and Metro Ethernet
supports manual configuration; for this bearing architecture specify Forum (MEF).
•A Reliable network protection PW-based emulation services •Service and VC/CP binding
mechanism, capable in various PTN including TDM services, Ethernet L2 implemented through ports or port+
network layers and topologies ; services, and ATM services. However, Virtual Local Area Network (VLAN)
•Different QoS guarantees for the functional requirements of a should be supported.
different services; peer-to-peer IP and MPLS service •Network management-based
•Sound OAM fault management and bearing model are still being studied. static configuration of Ethernet services
performance management; All service bearing in PTN is connection should be supported. Signalling based
•Label-based packet forwarding. oriented; a non connection-based dynamic configuration of Ethernet
Encapsulation, transport, and mechanism is not included in the services is optional.
processing of OAM messages are standard. •The PWE3 encapsulation and
independent of IP encapsulation and MPLS-TP PTN should meet the control word should follow RFC4448.
02 ZTE COMMUNICATIONS September 2010 Vol.8 No.3
Study of the China PTN Industry Standard and Its Key Issues Special Topic
(4) PTN support for ATM service including unidirectional/bidirectional PTN bottom network (WDM and OTN) is
emulation and transport is optional, and 1+1 path protection, bidirectional 1:1 or configured with protection, PTN
the PWE3 encapsulation and control 1:N (N>1) path protection, network protection should support a
word should follow RFC4717. PTN unidirectional/bidirectional 1+1 SNC/S hold off time, which can be set at 50 ms
supports both 1:1 and N:1 Virtual protection, and bidirectional 1:1 SNC/S or 100 ms.
Channel Connection/Virtual Path protection. At minimum, a bidirectional
Connection (VCC/VPC) encapsulation, 1:1 protection mechanism should be 2.4 PTN OAM Architecture and
as well as the establishment of supported. Functional Requirements
unidirectional and bidirectional •Ring protection within PTN PTN OAM includes the OAM
point-to-multipoint VCC/VPC. Inverse including Wrapping and Steering. At mechanism within PTN, service layer
Multiplexing over ATM (IMA) group minimum, one of these two protection OAM, and access link layer OAM.
processing is optional. mechanisms should be supported. The OAM mechanism within PTN is
MPLS-TP based PTN should meet (2) Access Link Protection of PTN divided into alarm-related OAM,
the following data forwarding and Other Networks performance-related OAM, and other
requirements: •1+1 or 1:N protection of TDM/ATM OAM. The OAM requirements of the VC,
(1) The MPLS-TP data forwarding access links; VP, and VS layers are shown in Table 1.
mechanism is a subset of the MPLS •Protection of GE/10GE access Proactive OAM periodically or
data forwarding system (RFC3031), links, eg. Link Aggregation (LAG). continuously reports fault and error
and should meet the transport (3) Dual-Homed Protection code performance, whereas
requirements specified in RFC5654. Protection within PTN and access link on-demand OAM is manaually initiated
MPLS-TP does not use hop-by-hop protection are combined for for the purposes of fault diagnosis and
forwarding based on IP addresses, and end-to-end service protection when positioning.
does not support the merged function the access link or PTN access node
of ECMP, PHP, and LSP. fails. This mechanism is still being 3 Key Issues Associated
(2) MPLS-TP label stacking should studied.
follow RFC3032 and RFC5462. The Protection within PTN should meet
with the China PTN
MPLS Time-to-Live (TTL) processing the following generic functional Industry Standard
mechanism should follow RFC3443. requirements: In the drafting of General Technical
(Both VC and VP Transmission (1) PTN protection switching should Requirements for Chinese Packet
Convergence (TC) and TTL should support triggering of link/node fault Transport Networks, selection of
adopt pipe models.) conditions and external commands technologies for the standard, PTN
(3) The range of PWs and LSP labels from the network management system, applications, and sustainable
should be 1-1,048,575 (eg. 220-1), as well as priority handling of various development of PTN were disputed.
from which 0-15 are reserved labels switching requests. The types of faults After extensive discussions, some
(used or reserved for PTN OAM. triggered include Signal Failures (SF), issues have been solved, but others
RFC3032 defines 4 reserved label intermediate node failures of both require follow-up study.
values). physical links and VP/VC, and Signal (1) MPLS-TP OAM Implementation
(4) MPLS-TP should support the Degrade (SD). External commands that Mechanism and Encapsulation Format
following label switching functions: are triggered result in network There are three available options:
Correct pushing of PWs and LSP management orders such as "locked T-MPLS G.8114, G-ACh[5]+Y.1731
labels by the source node, enabling protection", "forced switching", "manual OAM PDU[6], and IETF BFD expansion[7]
multiple PWs to be multiplexed into one switching", and“clear command” . +new OAM tools. All Chinese members
LSP; (2) Protection switching mode: This of the standardization group have
Correct popping of PWs and LSP supports both single-end and agreed to adopt G-ACh+Y.1731 OAM
labels by the sink node; dual-end switching types. It should PDU to preserve existing benefits of
Supporting Single-Segment Pseudo support configuration of "non-revertive Chinese operators and vendors, and to
Wires (SS-PW) and Multi-Segment operation" or "revertive operation" facilitate software upgrade in the future.
Pseudo Wires (MS-PW) switching. mode, as well as initiation of Wait to All members wish to promote their
SS-PW architecture should follow Restore (WTR) function and WTR time preference as the international
RFC3985. MS-PW switches PWs of setting. MPLS-TP standard. However, selecting
different LSPs into the same LSP. (3) Protection switching time: The G-ACh+Y.1731 OAM PDU as the
service disruption time caused by MPLS-TP standard is difficult because
2.3 PTN Network Protection Functional protection switching—except the group is led by non-Chinese giants
Requirements SD-triggered protection Cisco and Juniper.
PTN supports three protection types: switching—should not exceed 50ms (2) Implementation Mechanism of
(1) Protection within PTN when the delay time is set at 0. PTN Ring Protection
•Linear protection within PTN (4) Hold off time setting: When the IETF has approved ring protection
Special Topic Study of the China PTN Industry Standard and Its Key Issues
▼Table 1. OAM requirements in PTN industrial chain is expected to continue

developing.
Type Function VS OAM VP OAM VC OAM
References
Continuity Check and [1] 荆瑞泉. MPLS-TP 技术标准进展[R].专家技术报告.
Required Required Required 2009.
Connectivity Verification
JING Ruiquan. MPLS-TP technology standard
Remote Defect Indication Required Required Required progress [R]. TC6-2009-026Q Experts’ Technical
Fault Reports. 2009.
Management Alarm Indication Signal Not Applicable Required Required [2] 李芳.《分组传送网(PTN)总体技术要求》征求意见稿
Proactive
OAM LI Fang. Draft of general technical requirements for
Locking Required Required Required
Chinese packet transport networks for comments
Client Signal Fault Not Applicable Not Applicable Required [S]. 2009.
[3] ITU-T SG15 G.ptneq. Types and characteristics of
Performance Monitoring
Packet Loss
Optional Required Optional Packet Transport Network (PTN) equipment [S].
Measurement 2009.
[4] BOCCI M, BRYANT S, FROST D, et al. A framework
Loopback Required Required Required
for MPLS in transport networks [R].
Link Trace Not Applicable Required Required draft-ietf-mpls-tp-framework-10. 2010.
Fault Management and [5] IETF RFC 5586. MPLS generic associated channel
Location
Testing Optional Required Required [S]. 2009.
On [6] BUSI I, van HELVOORT H, HE J. MPLS-TP OAM
Demand Locking Optional Required Required based on Y.1731 [R]
OAM
draft-bhh-mpls-tp-oam-y1731-04. 2010.
Packet Loss
Measurement
Optional Required Required [7] FULIGNOLI A, BOUTROS S, VIGOUREUX M.
Performance Monitoring Proactive connection verification, continuity check
Packet Delay and remote defect indication for MPLS transport
Optional Required Required
Measurement profile [R]. draft-asm-mpls-tp-bfd-cc-cv-01.
2009.
Automatic Protection Switching Required Required Optional
[8] IETF RFC 5317. Joint Working Team (JWT) report on
Management/Signaling Control Path Not Applicable Required Optional MPLS architectural considerations for a transport
profile [S]. 2009.
Others Clock Optional Not Applicable Not Applicable [9] IEEE 802.1Qbf. PBB-TE infrastructure segment
protection [S]. 2009.
Operator's Custom Functions Not Applicable Optional Optional [10] ITU-T SG15 G.8132 (draft). T-MPLS shared
protection ring (TM-SPRing) [S]. 2007.
Functions Used for Testing Not Applicable Optional Optional
OAM: Operation, Administration and Maintenance VP: Virtual Path

VC: Virtual Channel VS: Virtual Section
Biographies
requirements, with implementation deal with PTN and MSTP mixed
Zhang Haiyi
mechanisms including MPLS-TE Fast networking, and interworking of two
Re-route (FRR) applied in ring networks. Such a scenario might Zhang Haiyi is director and
topology[8], IEEE multi-segment include PTN convergence ring senior engineer at the Transport
and Access Research
protection[9], and ITU-T Wrapping/ interworked with the MSTP access ring. Department of the Institute of
Steering ring protection[10]. In However, specifications of the related Communication Standards
consideration of Chinese vendor standard are still open and being Research, China Academy of
Telecom Research. She is also
products and operator applications, the debated. deputy leader of the transport
standard adopts ITU-T Wrapping/ (4) Requirements of PTN Supporting network working group of the
China Communications
Steering ring protection. Prior to IP/MPLS L3 Functions Standards Association (CCSA).
selection, three technologies were With the trend towards all-service She has long been engaged in researching optical
deeply analyzed in terms of service operation and Long Term Evolution transmission systems, WDM systems, SDH systems,
MSTP, Automatic Switched Optical Networks
configuration, bandwidth sharing, (LTE) mobile backhaul bearing, should (ASON), and telecom transport network
OAM, and cross-ring protection. The PTN support certain IP/MPLS L3 standardization. Zhang is also a technical consultant
to telecom operators. She has published 30
impacts on network operation and functions? If so, how should it academic papers.
maintenance were also taken into cooperate with the existing router
consideration. Follow-up network? These questions will be
Li Fang
standardization work will involve studied after operators’specific
specific mechanisms for requirements are made clear. This is an
Li Fang is a directior engineer at
point-to-multipoint service protection open issue in PTN standardization. the Institute of Communication
and cross-ring protection. PTN is developing, competing, and Standards Research, China
Academy of Telecom Research.
(3) PTN and Multiservice Transport converging with IP/MPLS technology. It Her research interests include
Platform (MSTP) Mixed Networking and has so far been applied in IP-based 3G PTN, mobile backhaul, IP and
optical network convergence,
Interworking Requirements mobile backhaul networks. The next and high-speed optical
The mission of PTN is to gradually two years will be key in the transmission. She has
replace SDH-based MSTP. Therefore, standardization and industrial published 20 academic papers.
in network deployment, operators must application of PTN technology. The PTN
Standardization Progress of Packet Transport Networks Special Topic
Yang Jian
Standardization Progress of
Packet Transport Networks
Yang Jian
(Standard Development and Industry Relations Department, Shenzhen 518057, P. R. China)
Abstract:
Global operators are in the process of constructing next generation IP packet-switched networks, with the view of moving towards full
IP-based networks. As data services increase, improvements in bandwidth, Quality of Service (QoS), Operation Administration and
Maintenance (OAM), and network reliability are required. Convergence of network data transmission and network communications
technology has been driven by converged IP services (including video, voice, and data), and unified multi-service load demand. Packet
Transport Networking (PTN) has come into being to meet these requirements, and PTN Standards are being developing rapidly.
T
here are two leading Packet MPLS-TP has won more support from connection-oriented and avoids the
Transport Network (PTN) vendors and operators than PBB-TE; flooding of broadcasting packets. This
technologies: Provider Backbone therefore, it is more likely to be enables connection configuration
Bridge Traffic Engineering deployed in large-scale future through network management system
(PBB-TE) and MPLS Transport Profile networks[1-4]. or control protocols so that carrier class
(MPLS-TP). The former adapts protection switching, OAM, QoS, and
Ethernet technology to carrier-class TE are achieved.
transport networks, while the latter is 1 PBB-TE PBB-TE uses IEEE 802.1ag as the
developed from transport networks and PBB-TE (IEEE 802.1Qay) is based on signaling protocol for PBB-TE
Multiprotocol Label Switching (MPLS). PBB (IEEE 802.1ah), which is also tunneling, and continuously monitors
PBB-TE is derived from Provider called MAC-in-MAC. Based on MAC tunnel states in the network. If the
Backbone Bridge (PBB) technology. stacking technology, it encapsulates working tunnel fails, services are
PBB-TE eliminates Media Access user MAC within the operator’ s MAC, automatically transferred to a
Control (MAC) address learning, enhancing Ethernet scalability and pre-established backup circuit. This
spanning tree protocol and flooding. It service security by using double increases network flexibility.
has connection-oriented features and encapsulation to isolate user traffic. The IEEE approved the establishment
improved network scalability. critical part of PBB involves the and authorization application of the
MPLS-TP is a connection-oriented introduction of a 24 bit Service Instance PBB-TE program, setting up the
packet-switched technology capable Tag (I-TAG) into the MAC-in-MAC 802.1Qay task group in March 2007.
of Traffic Engineering (TE) and encapsulation. The group began standardization of the
managing network resource use. The PBB-TE provides Ethernet with PBB-TE (IEEE 802.1Qay) data plane in
MPLS-TP data forwarding plane is a connection-oriented forwarding, which April 2007, and voted through
subset of MPLS defined by the Internet enables service providers to offer 802.1Qay Draft 1.1 in January 2008. In
Engineering Task Force (IETF). dedicated Ethernet links and to January 2009, IEEE introduced the draft
MPLS-TP has MPLS features, but also guarantee performance. In other words, version of 802.1Qay D5.0 and
meets transport requirements. It has a PBB-TE provides strict QoS for a Metro promoters voted on it. IEEE
reliable protection mechanism, Ethernet. Certain traditional Ethernet 802.1Qay-2009 was officially released
improved Quality of Service (QoS), and functions such as MAC address in August 2009.
enables scalable Operation learning, broadcasting, spanning tree Although it has been officially
Administration and Maintenance (OAM). protocol, and multicasting are released, IEEE 802.1Qay only specifies
In terms of industrialization, eliminated. PBB-TE is therefore the end-to-end protection of Traffic
Special Topic Standardization Progress of Packet Transport Networks
Yang Jian
Engineering Service Instance (TESI). In

MPLS Architecture
order to improve reliability and
IETF MPLS MPLS-TP
flexibility, several companies applied a
PBB-TE segment protection program
(a kind of local protection program) T-MPLS
Requirements
and, in a plenary session held in July
2008, debated whether local protection MPLS-TP
should be implemented in infrastructure G.8101v2
G.8110.1v2
or TESI. ITU-T T-MPLS G.8121v2
Revise to Align
The conclusion of that session was to G.8101
and Complete
G.8131v2
Copy Frame G.8110.1 G.8151v2
make infrastructure protection for Format and G.8121 G.81xx
available programs, and to make Forwarding G.8131 G.81yy
G.8151
improvements of TESI local protection Plane
as later required. After a year-long IETF: Internet Engineering Task Force MPLS: Multiprotocol Label Switching
Figure 1.▶
discussion, application of the PBB-TE ITU-T: International Telecommunication MPLS-TP: MPLS Transport Profile
Evolution of
segment protection program (program T-MPLS/MPLS-TP
Union’s Telecommunication T-MPLS: Transport MPLS
Standardization Sector
number IEEE 802.1Qbf) was finally standards.
approved at the plenary session in July
2009. The 802.1 working group by eliminating technical content that is cooperate to develop related
released the first draft of IEEE 802.1Qbf not connection-oriented, abandoning standards. ITU-T would provide
D0.0 and carried out the first vote in complex protocol families, and adding transport requirements to IETF; and to
September 2009. The discussion at the OAM and protection in the style of meet these requirements, would
plenary session held in November 2009 legacy transport networks. In 2006, the expand MPLS OAM, network
focused on overlapping protection ITU first approved three standard management, and control plane
groups. recommendations for T-MPLS protocols through the IETF’ s standard
As for the PBB-TE control plane, two architecture, interface, and equipment procedure. The technology was
technologies are available for functions and features. Following that, renamed MPLS-TP. IETF’ s MEAD
dynamically configuring PBB-TE the standard recommendations for team was responsible for defining
tunnels: Provider Link State Bridging OAM, protection, and network MPLS-TP. In July 2008, the IETF 72nd
(PLSB) and Generalized Multiprotocol management were successively meeting released MPLS-TP related
Label Switching (GMPLS). PLSB is proposed. In the ITU-T SG15 individual drafts on requirements,
802.1aq under the specification Q12+Q14 interim meeting held in framework, MPLS generic associated
process of the IEEE 802.1 working Stuttgart in September 2007, the IETF channel, network management, OAM
group, while GMPLS is a PBB-TE mission indicated that ITU-T’ s analysis and requirements, and
related standard being developed by T-MPLS standards conflicted with survivability framework.
the IETF Common Control and IETF’ s related standards. Reserved WP3 openly discussed
Measurement Plane (CCAMP) working bytes and reserved numbers of MPLS T-MPLS/MPLS-TP documents at the
group. GMPLS should coordinate with protocols, for example, were used February 2008 ITU-T SG15 plenary
the data plane standardization of the without any negotiation with IETF, even meeting. The WP3 chairman used
IEEE. Documents on the PBB-TE though IETF defined the core part of pages 7-10 of the powerpoint
control plane released by the IETF MPLS protocols. document provided by the JWT to
CCAMP working group include a few on At the February 2008 ITU-T SG15 eliminate bias and make clear the
the requirements, architecture, and plenary meeting held in Geneva, following: MPLS-TP standards would
control protocols of GMPLS controlling Groups Q12 and Q14 reported the be jointly developed by IETF and
Ethernet. debate on T-MPLS protocols. The ITU-T, as shown in Figure 1; ITU-T
meeting resulted in a Joint Working would stop work on T-MPLS standards;
Team (JWT) being established to the released T-MPLS standards would
2 MPLS-TP evaluate options for evolving MPLS remain unchanged; ITU-T would not
technology to meet requirements of the make any standard revision or engage
2.1 Development History transport network. The JWT comprised in new standard development until
The International Telecommunication the T-MPLS Ad Hoc group of ITU-T IETF’ s related standards became
Union’s Telecommunication and MPLS Interoperability Design stable; and the technology name would
Standardization Sector (ITU-T) SG15 (MEAD) team. The JWT chairmen were be changed to MPLS-TP.
started working on Transport MPLS Malcolm Betts, the Q12 rapporteur, and Drafts on MPLS-TP framework,
(T-MPLS) standardization as early as David Ward from IETF. After a series of network management, MPLS generic
2005. Based on MPLS and transport teleconferences, in April 2008, the JWT associated channel, and OAM that
architecture, T-MPLS simplifies MPLS decided that ITU-T and IETF would were submitted to the IETF 74th
Standardization Progress of Packet Transport Networks Special Topic
Yang Jian
This provides a list of MPLS-TP

G.8112 related Requests for Comments (RFC)
and standards being developed. It also
G.7712 G.8151 G.8152 shows the steps involved in RFC
development. The working group’ s last
G.8121 call is critical in the RFC release
procedure. The last call of a RFC
G.8110.1 usually requires two weeks, and may be
G.mpls-tp oam repeated once. In total, the last calls of
20 standards require about 80 weeks.
G.8131
Renewing the last call should be
avoided in order to save time.
Therefore, ITU-T experts are expected
G.8132 to submit informal comments in the
early stage of opening an MPLS-TP
▲Figure 2. Reference relations among ITU-T standards. document, and this can avoid or
significantly decrease comments from
meeting in March 2009 became the and representative of IETF, would be ITU-T in the formal review period of a
working group documents. responsible for the general coordination last call.
ITU-T SG15 Q9, Q10, Q12, and Q14 of MPLS-TP work, while the chairmen Figure 2 shows reference relations
held a joint meeting to discuss revision of the five working groups would among ITU-T standards.
of MPLS-TP standards in May 2009. At execute detailed working plans. A Figure 2 does not show G.8101
the meeting, editors provided the committee would be established to because the development of G.8101
revised versions of G.8110.1, G.8110.1 coordinate with ITU, with members depends on the completion of all the
amd1, G.8112, G.8121, G.8131, including the study subject chairmen, other standards, including the
G.8151, G.8101, the G.8132 draft, and WP3 chairman, the SG15 chairman of terminologies. Dotted lines in Figure 2
the G.mpls-tpoam draft. At the meeting ITU-T, as well as the working group indicate the initial version of G.8151,
it was agreed that standards would be chairmen and routing area chairman of which may not include the linear and
approved in the sequence G.8110.1, IETF. The committee members would ring protection models.
G.8101, G.7712, G.8112, G.8121, convene regularly to exchange The MPLS-TP standards to be
G.8131 and G.8151. information and promote MPLS-TP submitted for approval are divided into
standardization. Information arising three stages: G.8110.1, G.7712,
2.2 Progress of Standard Development from this committee would be provided G.8112; G.mpls-tp oam, G.8121,
to ITU-T participants through the G.8151; and G.8131, G.8132, G.8152.
2.2.1 ITU-T mailing list. The original JWT and Ad If a referenced RFC is approved, the
The second ITU SG15 meeting in the Hoc was to remain unchanged, but the corresponding ITU-T standard may
2008-2012 study period was held in Ad Hoc website would be updated to also be approved, and IETF should
Geneva between September 28 and reflect current work. check the corresponding ITU-T
October 9, 2009. The joint meeting of (2)TD167-WP3 "LS - New version of standard at the same stage. The
MPLS-TP Q9, Q10, Q12, and Q14 the MPLS-TP process document original versions of these standards
discussed IETF mail correspondence, available" have existed, and they must be in line
and also created new correspondence. This document aimed to explain the with developed RFCs. The current plan
(1)TD218-WP3 "LS: Restructure of normal IETF working procedure and involves approving the first-stage
MPLS-TP Work Forums in the IETF." interactions with ITU-T. It is helpful for ITU-T standards at the next SG15
IETF notified ITU-T that it would the ITU-T members to understand the plenary session, and these standards
dismiss the IETF MEAD team with the IETF procedure and to participate in will be discussed through mail and
purpose of conducting MPLS-TP IETF work according to the procedure. interim meetings before the session.
standardization work according to the This document is currently a working
normal IETF standardization procedure. group draft, and will be further 2.2.2 IETF
It would enable all IETF members to promoted. It is expected to be released At the IETF 76th meeting held in
participate in MPLS-TP standardization as a history RFC. The draft should be Hiroshima between November 8 and
work earlier. MPLS-TP related tasks updated according to TD218-WP3, 13, 2009, the MPLS-TP drafts were
would be undertaken by five related and ITU-T has provided some discussed by the MPLS working group.
IETF working groups. Discussions on comments and sent correspondence Since there were many MPLS-TP
MPLS-TP might be conducted through to IETF. related personal documents submitted
the IETF MPLS-TP mailing list. Adrian (3)TD200-WP3 "LS: Progress report for the meeting, the MPLS working
Farrel, the chairman of the routing area on MPLS-TP documents" group arranged two MPLS meeting
Special Topic Standardization Progress of Packet Transport Networks
Yang Jian
▼Table 1. MPLS-TP related RFCs and working group documents segments and one meeting segment of
the Pseudo Wire Emulation
Document State Document Name
Edge-to-Edge (PWE3) working group
RFC 5317: JWT Report on MPLS Architectural Considerations for a Transport Profile
to discuss them.
RFC 5462: EXP field renamed to Traffic Class field Table 1 lists MPLS-TP related IETF
RFC 5586: MPLS Generic Associated Channel RFCs and working group documents,
RFC 5654: MPLS-TP Requirements and Table 2 lists the individual
RFCs
RFC 5718: An Inband Data Communication Network For the MPLS Transport Profile documents submitted for the IETF 76th
RFC 5704: Uncoordinated Protocol Development Considered Harmful
meeting.
RFC 5860: Requirements for OAM in MPLS Transport Networks
RFC 5921: A Framework for MPLS in Transport Networks

3 Conclusion
draft-ietf-mpls-tp-nm-req: MPLS TP Network Management Requirements The emergence of IP services results in
draft-ietf-mpls-tp-cc-cv-rdi: Proactive Connection Verification,
Continuity Check and Remote Defect Indication for MPLS Transport Profile
IP based networks. Legacy networks
using Time Division Multiplexing (TDM)
draft-ietf-mpls-tp-ach-tlv: Definition of ACH TLV Structure
switching are gradually shrinking and
draft-ietf-mpls-tp-survive-fwk: Multiprotocol Label Switching Transport Profile Survivability Framework
being replaced by packet networks.
draft-ietf-mpls-tp-nm-framework: MPLS-TP Networks Management Framework
Packet transport networking is an
draft-ietf-mpls-tp-oam-framework: MPLS-TP OAM Framework and Overview
inevitable outcome of IP based
draft-ietf-mpls-tp-oam-analysis: MPLS-TP OAM Analysis
networks. The progress of its
draft-ietf-mpls-tp-rosetta-stone: MPLS-TP Rosetta Stone
standardization will significantly impact
MPLS Working Group Drafts draft-ietf-mpls-tp-linear-protection: MPLS-TP Linear Protection
the development of the entire industrial
draft-ietf-mpls-tp-identifiers: MPLS-TP Identifiers
chain.
draft-ietf-mpls-tp-process:
IETF Multi-Protocol Label Switching (MPLS) Transport Profile (MPLS-TP) Document Process
draft-ietf-mpls-tp-fault: MPLS-TP Fault OAM Refernces

draft-ietf-mpls-tp-data-plane: MPLS Transport Profile Data Plane Architecture [1] IEEE 802.1Qay. IEEE standards for local and
metropolitan area networks: virtual bridged local
draft-ietf-mpls-tp-on-demand-cv: area networks, amendment 7: provider backbone
MPLS on-Demand Connectivity Verification, Route Tracing and Adjacency Verification
bridge traffic engineering [S]. 2009.
draft-ietf-mpls-tp-loss-delay: Packet Loss and Delay Measurement for the MPLS Transport Profile [2] Report of working party 3/15, transport network
structures [R]. T09-SG15-081201-TD-WP3-0080
draft-ietf-mpls-tp-lsp-ping-bfd-procedures: LSP-Ping and BFD Encapsulation over ACH
MSW-E (Geneva meeting). 2008.
draft-ietf-ccamp-oam-configuration-fwk: [3] Report of working party 3/15, transport network
CCAMP Working OAM Configuration Framework and Requirements for GMPLS RSVP-TE structures [R]. T09-SG15-090928-TD-WP3-0212
Group Draft MSW-E (Geneva meeting). 2009.
draft-ietf-ccamp-mpls-tp-cp-framework: MPLS-TP Control Plane Framework
[4] IETF RFC 5654.MPLS-TP requirements [S]. 2009.
PWE3 Working Group Draft draft-ietf-pwe3-mpls-transport: Application of Ethernet Pseudowires to MPLS Transport Networks
OPSA Working Group Draft draft-ietf-opsawg-mpls-tp-oam-def: The OAM Acronym Soup
▼Table 2. Individual documents submitted for the 76th IETF meeting

Document State Document Name
draft-koike-ietf-mpls-tp-oam-maintenance-points-00
draft-sprecher-mpls-tp-oam-analysis-07
draft-asm-mpls-tp-bfd-cc-cv-01
draft-nitinb-mpls-tp-lsp-ping-extensions-00
draft-swallow-mpls-tp-identifiers-02
draft-sfv-mpls-tp-fault-00
draft-frost-mpls-tp-loss-delay-00
draft-sprecher-mpls-tp-survive-fwk-02 Biography
draft-weingarten-mpls-tp-linear-protection-04 Yang Jian
Individual Documents
draft-weingarten-mpls-tp-ring-protection-01
draft-umansky-mpls-tp-ring-protection-switching-00
draft-liu-mpls-tp-ring-protection-00 Yang Jian is an engineer at the
draft-jiang--mpls-tp-ring-fd-00 Standards Development and
Industry Relations Department
draft-he-mpls-tp-csf-01 of ZTE Corporation. Her
draft-zhl-mpls-tp-sd-01 research focuses on the
standards for packet transport
draft-flh-mpls-tp-oam-diagnostic-test-00
networks.
draft-dai-mpls-tp-lock-instruct-00
draft-chen-mpls-tp-nm-discovery-req-00
Several Issues in the Development of Packet Transport Networks Special Topic
Jing Ruiquan
Several Issues in the Development

of Packet Transport Networks
Jing Ruiquan
(China Telecom Corporation Limited Beijing Research Institute, Beijing 100035, P. R. China)
Abstract:
Several key issues affect the development and standardization of Packet Transport Networks (PTN) and Multiprotocol Label Switching
Transport Profile (MPLS-TP). These include the end-to-end Quality of Service (QoS) mechanism, layered network architecture,
introduction of layer-3 functions, and data-plane loopback functions. This paper introduces several views on the construction and
maintenance of PTN, and on the requirements of PTN services. After an analysis of Traffic Engineering (TE) based on Multiprotocol Label
Switching (MPLS) and Differentiated Service (DiffServ), a service-oriented end-to-end QoS guarantee mechanism is proposed. An
alternative to introducing layer-3 functions into PTN is also proposed, based on PTN layered architecture defined in the available
MPLS-TP standards and drafts. Requirements of data-plane loopback functions are discussed in conclusion.
T
here is no standard definition for reflects a history of competition and QoS in PTN, with the goal of creating a
Packet Transport Networks integration of the transport and data service-oriented end-to-end QoS
(PTN). Broadly speaking, PTN fields. MPLS-TP is an achievement guarantee[7-10].
can be defined as any transport preceded by years of competition and
network that is based on packet negotiation between the Internet 1.1TE
switching and meets certain Operation Engineering Task Force (IETF) and According to IETF’ s definition,
Administration and Maintenance International Telecommunication MPLS-TP must support TE so that
(OAM), protection, and network Union’ s Telecommunication network resources can be controlled.
management requirements. Specific Standardization Sector (ITU-T). It can The purpose of TE is to optimize
packet-switched technologies include be seen as a balanced exploitation of resource use for effective and reliable
Multiprotocol Label Switching (MPLS), the benefits of transport and data fields. network operation. An important
Transport MPLS/MPLS Transport Profile IETF currently leads the development of element of TE is the Constraint-Based
(T-MPLS/MPLS-TP), Ethernet, MPLS-TP standards, while ITU-T Routing (CBR) mechanism. In an
Backbone Bridge Traffic Engineering SG15 has a lesser hand and mainly IP/MPLS network, TE is generally
(PBB-TE), and Resilient Packet Ring relies on the participation of enterprises implemented by MPLS-TE, and has
(RPR). Over the past two years, both and individual expert members[1-6]. two important roles:
T-MPLS and PBB-TE have been (1) Making Service Routing
viewed as the leading PTN Controllable
technologies. However, as support for 1 QoS in PTN A PTN service is first encapsulated
PBB-TE from vendors and operators QoS provides predictable service by the Pseudo Wire (PW), and then
wanes, China has come to view quality in terms of packet loss ratio, multiplexed to the Label Switched Path
T-MPLS/MPLS-TP as PTN. Therefore, delay, jitter, and bandwidth during (LSP). The LSP is established through
PTNs described in this paper are based network communication. QoS in PTN network management or the control
on T-MPLS/MPLS-TP. includes stream classification, labeling, plane, and LSP routing implemented by
Transition from T-MPLS to MPLS-TP rate limitation, bandwidth guarantee, either of these is controllable.
traffic shaping, and scheduling (2) Making Service Bandwidth
This work was funded by the National High
strategy. MPLS Traffic Engineering (TE) Controllable
( program) under Grant and Differentiated Service (DiffServ) PTN can bear E1 emulation services
No. 2007AA01Z2A4. mechanisms are used to implement and Ethernet services. The bandwidth
Special Topic Several Issues in the Development of Packet Transport Networks
Jing Ruiquan
of an E1 emulation service must be (EF) and Assured Forwarding (AF). between LSP and BA:
fixed and controllable, having high (1) EF EXP-inferred-PSC LSP (E-LSP) and
priority, no packet loss, and low delay. EF PHB traffic is unaffected by any Label-Only-Inferred-PSC LSP
Ethernet services can be divided into other PHB traffic, and this ensures (L-LSP).
Constant Bit Rate (CBR) services and packets are forwarded at the quickest (1) E-LSP
Variable Bit Rate (VBR) services. A CBR possible rate. Similar to legacy leased E-LSP uses the MPLS-labeled EXP
has the same basic requirements and lines, EF PHB can guarantee bandwidth field to designate multiple BAs to one
controllability as E1 emulation services. service with low packet loss, low delay, LSP. The MPLS-labeled EXP field
A VBR service uses Committed and low jitter. CIR is the only bandwidth represents the PHB of a packet. Up to
Information Rate (CIR) and Excess parameter of an EF-based service; EIR eight BAs can be mapped into the EXP
Information Rate (EIR) to control is equal to 0 and traffic greater than CIR field; that is, an E-LSP supports a
bandwidth. An operator only is discarded. EF can be applied to E1 maximum of eight service levels.
guarantees user bandwidth not emulation and CBR Ethernet services, (2) L-LSP
exceeding the CIR, and may discard and should follow RFC3246. L-LSP designates one LSP to one
EIR traffic when the network is (2) AF BA, and uses EXP to represent the
congested. In this way, network AF provides four levels of packet packet discarding priority. An L-LSP
bandwidth is controlled. forwarding, each level with three can only support one service level.
More specifically, the operator first discarding priorities. The level of a MPLS equipment switches label
configures the CIR of the PW (eg. the service is determined by PTN values per hop, but the management of
service) as well as the LSP, and sets equipment—which configures mapping between label and PHB is
the conditions for controllable forwarding resources (such as buffer difficult. Therefore, compared to
connection. The sum of all the PW CIRs and bandwidth) at different levels, and L-LSP, E-LSP is more easily controlled
in one LSP should not exceed the CIR accounts for the discarding priority. If a because it can determine in advance
of the LSP; the sum of all the LSP CIRs service is not congested, AF service the mapping relationship between the
in one link should not exceed the CIR of performances at different levels are EXP field and PHB for every packet in
the link. Under these conditions, the equal. However, if the service is the network. PTN equipment currently
network can satisfy the CIR bandwidth congested, packet loss occurs on all uses E-LSP.
requirements of all services. Since VBR AF levels, and the degree of packet
service involves bursts (eg. the EIR loss correlates with the service level. AF 1.4 End-to-End QoS Implementation in
part), network congestion cannot be should follow RFC2597. PTN
completely avoided using TE. DiffServ TE and DiffServ supported by MPLS
is an effective solution for ensuring CIR 1.3 DiffServ Supported by MPLS help PTN guarantee service-oriented
bandwidth in such conditions. MPLS-TP based PTN must use the end-to-end QoS. MPLS TE is used to
MPLS DiffServ mechanism defined in control service routing and bandwidth
1.2 DiffServ RFC3270. in order to avoid network congestion
DiffServ is derived from Integrated After an IP packet has been caused by unbalanced loading. Once
Service (IntServ), and aims to provide encapsulated through MPLS, the core the network is congested by burst
differentiated service levels for Internet router cannot locate DSCP. service or network protection, DiffServ
traffic. Compared with IntServ, DiffServ Accordingly, IETF has proposed a supported by MPLS is used to
has a simpler control system with DiffServ-supported MPLS that can map guarantee CIR.
coarser granularity. It is used to control multiple Behavior Aggregates (BA) of Table 1 classifies service levels for
each QoS type after streaming DiffServ to a MPLS LSP and forward the E-LSP. Peak Information Rate
convergence. IntServ, on the other traffic on the LSP according to the BA (PIR)=CIR+EIR; and under this
hand, aims to control each individual PHB. There are two mapping modes condition, the EXP value of LSP and PW
stream. DiffServ is therefore scalable,
▼Table 1. PTN service priority instances
and capable of QoS in large-scale
networks. Service Serive (PW)
Experimental Field
Default
Service Type Instances
Levels Bandwidth Attribute Hop-by-Hop
DiffServ classifies IP streams into
Control/Network Management
different types at the edge of its 3 CIR=PIR 111 EF
Messages/E1 Emulation
domain, and assigns one Differentiated 2 CIR=PIR 110 EF Voice/Video
Services Code Point (DSCP) to each
101 (less than CIR)
stream type. The core router in the 1 CIR, PIR AF VPN/Private Ethernet
domain checks the DSCP value, and 100 (more than CIR but less than PIR)
dispatches packet forwarding 0 CIR=0, PIR 000 DF Best Effort Data Services
according to different types of Per-Hop AF: Assured Forwarding EF: Expedited Forwarding VPN: Virtual Private Network
Behavior (PHB). IETF has defined two CIR: Committed Information Rate PIR: Peak Information Rate
kinds of PHB: Expediated Forwarding DF: Default Forwarding PW: Pseudo Wire
Several Issues in the Development of Packet Transport Networks Special Topic
Jing Ruiquan
in a data frame are the same one. should also support related functions of but it is still doubtful whether it can meet
Both E1 emulation services and CBR the service layer and section layer. the bearing requirements of future LTE
Ethernet services (such as voice and Such functions include OAM of the systems.
video) adopt EF PHB with a setting of Ethernet service layer (specified in LTE systems have different bearing
CIR=PIR. IEEE802.1ag and Y.1731), OAM of the requirements from 3G networks
Burst-type services such as virtual Ethernet link layer (specified in IEEE because they require interconnection
private networks and private Ethernet 802.3ah), overhead handling of SDH between base stations (X2 interface)
lines use AF PHB. To ensure CIR services and links, and protection. and multi-homing from base station to
bandwidth of burst-type services, Current PTN equipment uses PW to Service Gateway (SGW). There are two
service streams must be measured, support various emulation services but ways to satisfy the bearing
shaped, and labeled at the network does not support IP/MPLS services requirements of LTE: end-to-end
ingress according to bandwidth through LSP. IP/MPLS service router networking, and L3+L2
parameters, and Two Rate Three Color implemented by Ethernet PW emulation networking (with L3 networking for the
Marker (trTCM) must be supported. is highly transparent, but also core layer and L2 networking for the
Moreover, the EXP value of the data inefficient—especially for short packets convergence and access layer). Since
frame is set from the mapping because Ethernet frame headers need end-to-end router networking still has
relationship, which is used by the LSP to be transported. IP/MPLS problems with network scalability,
follow-up nodes to select suitable PHB. implemented by TDM PW emulation has manageability, and controllability,
As for ordinary data services, CIR is high requirements on network L3+L2 networking has gained
set at 0, PIR is set, and Default performance, and may increase recognition and support. The core layer
Forwarding (DF) is used. equipment cost. IP/MPLS implemented in this solution can be implemented by
Even if the network is congested, the by service LSP can avoid the router networking or by introducing L3
service bandwidth of both EF PHB and abovementioned problems, but has functions into PTN.
CIR part of AF PHB traffic is always poor service transparency and is L3 functions mainly include IP routing
guaranteed. Ordinary data services are possibly required to handle part of the and forwarding, L3 MPLS Virtual Private
either discarded or weighed with AF L3 protocol. Therefore, service Network (VPN), and L3 multicasting. IP
PHB traffic so that certain bandwidth transparency, transport efficiency, and traffic and multicast, with uncertain
can be obtained during the congestion. cost must be taken into account when traffic bandwidth and routing, cannot
selecting IP/MPLS service provide strict QoS guarantee. If these
implementation technology. two services are introduced into PTN,
2 Layered Architecture In addition, current PTN equipment they should be set as the lowest level
of PTN only supports Single-Segment Pseudo services in order to avoid any impact on
IETF RFC5654 divides an MPLS-TP Wires (SS-PW); that is, where the L2 services. L3 MPLS VPN can use
system into transport service layer, source & sink of PW coincides with that MPLS TE and DiffServ mechanisms to
transport path layer, and section layer. of LSP. SS-PW cannot converge PWs guarantee QoS. MPLS VPN can also
The transport service layer can be PW borne by multiple LSPs, and requires support L3 multicasting with QoS
or service LSP, similar to Synchronous PTN equipment to have high LSP guarantee.
Digital Hierarchy (SDH) Virtual Channel capacity. Moreover, current PTN For PTN networks, L3 services with
(VC)-12. PW is used to offer emulation equipment with only end-to-end LSP QoS requirements can be offered by L3
services such as Time Division protection cannot cope with multi-point MPLS VPN, while L3 services with no
Multiplexing (TDM), Ethernet, and faults. Multi-Segment Pseudo Wires QoS requirements may be implemented
Asynchronous Transfer Mode (ATM) (MS-PW) can be introduced to solve directly through IP routing and
services. Service LSP is used to offer problems caused by SS-PW, and forwarding.
network-layer IP and MPLS services. thereby improve PTN scalability. IETF
The transport path layer refers to the has listed MS-PW as optional for
LSP layer, similar to SDH VC-4. The MPLS-TP.
4 Data-Plane Loopback
section layer is used to converge Functions
information from the transport service Current PTN equipment only supports
layer and transport path layer between
3 The Support of PTN on OAM Loopback (LB). OAM LB can be
two adjacent MPLS-TP nodes. The L3 Functions and Services used to verify bidirectional
section layer can be implemented by Currently, PTN mostly offers L2 interconnection between the source
either MPLS-TP or technologies such services, including E1/ATM emulation and sink maintenance endpoints, and
as SDH, Ethernet, or Optical Transport services and E-Line/E-LAN/E-Tree to check for inter-node/intra-node
Network (OTN). With a layered services. It is primarily applied in 3G faults. However, it cannot locate the
architecture, PTN can achieve and Long Term Evolution (LTE) mobile specific fault positions. As shown in
scalability similar to SDH/OTN. backhaul networks. PTN can satisfy the Figure 1, if there is a fault on the
Besides 3-layer MPLS-TP, PTN bearing requirements of 3G networks, PE2-PE3 link, OAM LB cannot determine
Special Topic Several Issues in the Development of Packet Transport Networks
Jing Ruiquan
quickly PTN moves from its introductory

PE1 PE2 PE3 stage into large-scale application.
(1) (2) (1) (2) (3) (1) (2) (3)
References
Tester [1] IETF RFC 5654. Requirements of an MPLS transport
• • profile [S]. 2009.
[2] IETF RFC 3270. Multiprotocol Label Switching
MEP MIP MIP MIP MIP MEP
(MPLS) support of differentiated services [S]. 2002.
[3] IETF RFC 3985. Pseudo Wire Emulation
Edge-to-Edge(PWE3) architecture [S]. 2005.
(1): Remote Loopback/Ingress Loopback [4] IETF RFC 5254. Requirements for multi-segment
Data-Plane Loopback (2): Local Loopback/Egress Loopback Pseudo Wire Emulation Edge-to-Edge (PWE3) [S].
(3): Fiber/Client Loopback 2008.
[5] IETF RFC 5659. An architecture for multi-segment
pseudo wire emulation edge-to-edge [S]. 2009.
MEP: Maintenance End Point MIP: Maintenance Intermediate Point PE: Provider Edge Router
[6] ITU-T Recommendation G.805(11/95). Generic
functional architecture of transport networks [S].
▲Figure 1. Data-Plane Loopback. 1995.
[7] BOCCI M, BRYANT S, LEVRAU L. A framework for
whether the fault is in PE3 or on the link. IETF and ITU-T are discussing the MPLS in transport networks [R].
draft-ietf-mpls-tp-framework-07 (work in
If PTN equipment can support standardization of data-plane progress). 2009.
SDH-like data-plane LB; that is, loopback functions. [8] BUSI I, ALLAN D. MPLS-TP OAM framework [R].
draft-ietf-mpls-tp-oam-framework-04 (work in
service LB, accurate fault positioning progress). 2009.
can be implemented through LB of [9] VIGOUREUX M, WARD D, BETTS M. Requirements
different points. 5 Conclusion for OAM in MPLS transport networks [R].
draft-ietf-mpls-tp-oam-requirements-04(work in
Similar to SDH, PTN data-plane PTN is the optimal solution for evolving progress). 2009.
loopbacks include Remote/Ingress Multiservice Transport Platform (MSTP) [10] IETF RFC 2698. A two rate three color marker [S].
1999.
Loopback, Local/Egress Loopback, networks bearing 2G mobile backhaul.
and Fiber/Client Loopback. Beside for The goal is to meet requirements of
fault locating, fiber loopback can be high-quality services such as 3G Biography
used to perform single-end service mobile backhaul and enterprise private Jing Ruiquan
performance tests on bidirectional time line services. From 2008 to 2009,
Jing Ruiquan is a senior
delay, packet loss ratio, and China’ s three dominant telecom engineer of China Telecom
throughout. It also supports testing in operators trialled PTN bearing 3G Corporation, Beijing Research
live networks. mobile backhaul in their live networks. Institute, and is a member of the
information communication
Supporting both remote and local They pushed forward the maturing and expert team of the National High
OAM loopbacks, PTN can implement commercialization of PTN products. Technology Research and
Development Program of China
functions similar to remote and local China Mobile began large-scale (
“863” program). His research
loopbacks on the data plane. Therefore, purchasing of MPLS-TP based PTN interests include optical
fiber loopback should be implemented equipment in October 2009, marking a transport networks, packet
transport networks, and IP RAN bearing technology.
in PTN to accurately locate faults, and turning point in the industrialization of He has long been engaged in planning, construction,
for single-end testing. Further research PTN. International MPLS-TP standards and technical research of optical transport networks,
and has participated in many SDH, MSTP, and ASON
is still needed to determine whether have attracted much attention in 2010, testing tasks. He has published more than 20
PTN should support data-plane remote and the stability of international academic papers.
and local loopback functions. Currently, MPLS-TP standards will determine how
Roundup
•• ••• • •• ••• ••• ••• • •• ••• • •• ••• ••• ••• • •• ••• • •• ••• ••• ••• • •• •••
•• ••• • •• ••• ••• ••• • •• ••• • •• ••• ••• ••• • •• ••• • •• ••• ••• ••• • •• •••
ZTE France Awarded Best Investor Prize 2010
•• ••• • •• ••• ••• ••• • •• ••• • •• ••• ••• ••• • •• ••• • •• ••• ••• ••• • •• •••
•• ••• • •• ••• ••• ••• • •• ••• • •• ••• ••• ••• • •• ••• • •• ••• ••• ••• • •• •••
•• ••• • •• ••• ••• ••• • •• ••• • •• ••• ••• ••• • •• ••• • •• ••• ••• ••• • •• •••
•• ••• • •• ••• ••• ••• • •• ••• • •• ••• ••• ••• • •• ••• • •• ••• ••• ••• • •• •••
ZTE Corporation has been awarded the high-tech industry. Eastern Europe and America.
the Best Investor Prize 2010 by The "We are proud to receive this award The award ceremony took place on
Greater Paris Investment Agency, from the Greater Paris Investment June 22nd in the Chamber of
recognizing it for its contribution to the Agency. ZTE is committed to growing Commerce in Paris and was given
Greater Paris region. its local office in France and making a personally by the secretary of state for
ZTE’s headquarters for Europe and contribution to the local community external trade, Ms. Anne-Marie Idrac.
North America is based in Paris, and this award recognises our In addition to ZTE, two companies in
helping to position the area as a achievements," said Mr. Zhu Yun, other industries were awarded the
strategic global business centre for President for ZTE Western Europe, Best Investor Price. (ZTE Corporation)
Features of PBB-TE Architecture and GMPLS Control Technology Special Topic
Wei Jianwen, Xie Rui, Jin Yaohui
Features of PBB-TE Architecture

and GMPLS Control Technology
Wei Jianwen1, Xie Rui 2, Jin Yaohui 1,2
(1. The state key lab of advanced optical communication systems & networks, Shanghai Jiao Tong University, Shanghai 200030, P. R. China;
2. Network & Information Center of Shanghai Jiao Tong University, Shanghai 200240, P. R. China)
Abstract:
Time Division Multiplexing (TDM) transport networks are evolving to packet-oriented, and a variety of carrier-class packet transport
technologies have emerged. Provider Backbone Bridge with Traffic Engineering (PBB-TE) is a connection-oriented packet transport
technology that provides good scalability and manageability, and guarantees Quality of Service (QoS). Generalized Multi-Protocol Label
Switching (GMPLS) is a mature transport network control plane technology that supports multiple data planes with different switching
granularity. GMPLS-controlled PBB-TE is a promising solution for Packet Transport Networks (PTN).
B
ackbone Bridge Traffic LAN Tag (C-Tag) domain based on the from customer VLAN, and allows
Engineering (PBB-TE) is a frame structure of 802.1 Ethernet is multiple customer VLAN services to be
connection-oriented packet introduced. C-Tag contains a 12-bit transported through the same provider
transport technology with good Customer Virtual LAN ID (C-VID) and a VLAN.
scalability and end-to-end QoS 3-bit Customer Product ID (C-PID). Limited by the length of the S-VID,
support. It contains an Operation, The C-VID indicates which VLAN the PBN supports up to 4,096 services. It
Administration and Maintenance (OAM) source host belongs to, and C-PID forwards frames in the format
mechanism on the data plane. This shows the service type of a frame. In a <C-DA+S-VID>, and learns customer
enhances the reliability and 802.1Q system, physical networks can Media Access Control (MAC)
manageability of telecom networks. support up to 4,096 VLANs, and traffic addresses. Each PBN node therefore
PBB-TE is expected to be the in different VLANs is separated. maintains a large forwarding table. With
preferred solution to metro Packet Depending on the service type these limitations, PBN fails to meet the
Transport Networks (PTNs). indicated by C-PID, the 802.1Q scalability requirements of telecom
network bridge can offer differentiated networks.
services.
1 Features of PBB-TE 1.1.3 802.1ah Provider Backbone
Architecture 1.1.2 802.1ad Provider Bridge (PB) Bridge (PBB)
IEEE 802.1ad PB[2] is the first In IEEE 802.1ah PBB[3] , a Provider
1.1 Evolution of IEEE Ethernet provider-oriented Ethernet bridge Backbone Bridge Network (PBBN)
technology. In PB, a Service VLAN Tag based on PBN is established to
1.1.1 802.1Q Virtual Local Area (S-Tag) domain is added to the 802.1Q improve PB scalability. A PBB frame
Network (VLAN) frame structure for service providers. has a provider frame header <B-DA,
In IEEE 802.1Q[1], a Customer Virtual This domain contains 12-bit provider B-SA, B-TAG, I-TAG>, which is
VLAN identifiers (S-VID) and a 3-bit absent from the PB frame. The PBBN
This work was funded by the National Basic
C-PID. An IEEE 802.1ad bridge edge node is responsible for adding
Research Program of China “973”
( Program) network is called Provider Bridge and deleting the provider frame header.
under Grant No. 2010CB328205, the National Network (PBN). As shown in Figure 1, In the header, B-DA and B-SA are the
Natural Science Foundation of China under
an S-Tag is either assigned or MAC addresses of the PBBN entry and
Grant No. 60825103, and the National Key
Technology R&D Program under Grant No. removed at the entry node of a PBN. exit nodes; the B-TAG containing a
2008BAH37B03. The S-Tag separates provider VLAN 12-bit B-VID indicates a PBBN
Special Topic Features of PBB-TE Architecture and GMPLS Control Technology
C-DA C-SA C-VID=100 C-DA C-SA C-VID=100 S-VID=100 C-DA C-SA C-VID=100
classification and control plane
technology are not developed enough.
These weaknesses are expected to be
solved during the PBB-TE
IEEE 802.1Q IEEE 802.1Q standardization process. PBB-TE
reduces MAN maintenance costs in the
long run, and some operators have
IEEE 802.1ad
already tried deploying PBB-TE
C-DA C-SA C-VID=100 C-DA C-SA C-VID=100 S-VID=101 C-DA C-SA C-VID=100 systems[6].
VLAN: Virtual Local Area Network
1.3 PBB-TE Equipment Interconnection
▲Figure 1. Provider VLAN is separated from customer VLAN in 802.1ad. Test in Shanghai Jiao Tong
University Backbone Network
spanning tree or a transport path; and having definite TE attributes and QoS A cost-effective carrier-class packet
the I-Tag with 24-bit I-SID represents guarantee. transport technology, PBB-TE can be
the number of services. A PBBN can (3) OAM applied not only in simple MANs, but
support up to 16 million services. The Using a Connectivity Fault also in complex scenarios (such as
PBBN core node forwards frames in the Management (CFM)[5] OAM mechanism, data centers or R&D backbone
format <B-DA+B-VID>, and only its PBB-TE can provide carrier-class networks with intensive services and a
edge nodes are required to learn OAM without assistance from other large number of nodes). The Network
customer MAC addresses. Therefore, layers. Center of Shanghai Jiao Tong
the forwarding table of the core node is (4) End-to-End Path Protection University has attempted to apply
greatly reduced. PBB-TE provides point-to-point and PBB-TE into the campus backbone
In terms of the number of services point-to-multipoint ESP with 1:1 path network, and the success of this test will
and nodes supported, PBB is the first protection. A protection path can be verify the interoperability of PBB-TE
bridge technology to meet the built while simultaneously building a and MPLS equipment.
requirements of telecom networks. working path, and by pre-configuring In the campus backbone network
However, PBB still lacks Traffic the protection path, QoS identical to the shown in Figure 3, the core network
Engineering (TE) and OAM features. working path can be achieved. PBB-TE consists of a series of IP/MPLS routers
path fault diagnosis and protection interconnected by 10GE or GE routers.
1.2 Features of 802.1Qay PBB-TE triggering are all completed on the data The convergence network consists of IP
A product of PBB and telecom network plane, and protection switching time routers. The campus network offers
features, IEEE 802.1Qay PBB-TE[4] is a can reach 50 ms. services such as on-demand and
type of connection-oriented packet (5) Multi-Service Bearing multicast intra-campus video, email,
transport technology. PBB-TE network PBB-TE can bear L2 and L3 File Transfer Protocol (FTP), and P2P
architecture, as shown in Figure 2, has services, and also supports TDM file sharing. To provide better service,
the following features: services. However, compared to the network uses MPLS-TE technology
(1) Scalability Transport Multiprotocol Label Switching in a number of areas. A specific
On the data plane, PBB-TE has the (T-MPLS), PBB-TE has weaker service-oriented MPLS Virtual Private
same MAC-in-MAC frame structure as Multipoint-to-Multipoint (MP2MP) Network (VPN) has also been created
PBB. The core node of a PBB-TE service support, and its QoS to provide university administration with
network forwards frames in the format
<B-DA, B-VID>. PBB-TE inherits the Payload
PBB-TE Network
S-VID
strengths of PBB in supporting a large C-SA Management ：Working Path
System
number of services and separating C-DA ：Protection Path
I-SID
provider and customer addresses. B-VID
(2) Connection-Orientation and QoS B-SA
Guarantee B-DA
PBB-TE discards the spanning tree

Q-in-Q Q-in-Q
protocol and source address learning
mechanism, as well as the frames of
unknown addresses. Ethernet Switched
PBB-TE
Path (ESP) is used for transport
Customer MAC Customer MAC
services and should be established by Address Learning Address Learning
the control plane or management MAC: Media Access Control
PBB-TE: Backbone Bridge Traffic Engineering
system. PBB-TE is therefore
connection-oriented with each ESP ▲Figure 2. PBB-TE architecture.
Features of PBB-TE Architecture and GMPLS Control Technology Special Topic
A new label format <B-DA, B-VID>

is added in GELS, which corresponds
：IP/MPLS Core Router
to the forwarding table entry of the
：IP Convergence Router
PBB-TE node. On the data plane,
different nodes on the same path
correspond to the same forwarding
：MPLS Edge Router
table entry; therefore, the nodes should
：PBB-TE Bridge be assigned the same label. The
PBB-TE label is global for the whole
：10 GE Link
network.
：1 GE Link
(3) Traffic Parameters
MPLS ：Multiprotocol Label
Switching GELS uses four bandwidth
PBB-TE ：Backbone Bridge parameters[8]: Committed Information
Traffic Engineering
Rate (CIR), Committed Burst Size
(CBS), Excess Information Rate (EIR),
and Excess Burst Size (EBS).
(4) Route and Path Computation
GMPLS does not limit path selection
Network that Participates in methods. Therefore, it allows
PBB-TE Interconnected Test computation and selection of any path.
Open Shortest Path First with Traffic
Engineering Extensions (OSPF-TE) and
▲Figure 3. PBB-TE Interconnected Testing System in the Backbone Network Intermediate System to Intermediate
of Shanghai Jiao Tong University. System Extensions for Traffic
Engineering (IS-IS-TE) can still be
a reliable and confidential network industry has generally agreed to use used to release routing messages of
platform. GMPLS as the PBB-TE control plane the PBB-TE data plane. Because of
In the test, the customer MAC frame technology. GMPLS extends the labeled or unlabeled data plane ports,
was encapsulated as the meaning of label and label exchange in routing messages need not carry MAC
MAC-in-MAC frame after the bridge MPLS, and reuses part of the MPLS addresses of ports.
convergence of PBB-TE. The frame protocol. GMPLS functions include (5) Link Management
was then encapsulated as an MPLS signaling, routing, path selection, and GMPLS Link Management Protocol
packet at the MPLS edge node, and link management. With extensions, (LMP) and PBB-TE CFM have some
transported by the MPLS core network. GMPLS may support data planes such overlapping functions. For example,
The frame arrives at the host through as SONET/SDH, Optical Transport both can implement neighbor
the MPLS edge router and PBB-TE Network (OTN), and Wavelength discovery, fault diagnosis, fault
bridge at the entry of the MPLS core Division Multiplexing (WDM). confirmation, and fault positioning. CFM
network. The PBB-TE bridge is a can work independently without
PBB-TE edge bridge with convergence 2.1 GMPLS Controlled of Ethernet support from other layers, while LMP
capability. Label Switching (GELS) can allocate numbered/unnumbered
The test will be deemed successful IETF is a leading promoter of interface IDs automatically. CFM and
when a PBB-TE convergence network standardization of GMPLS-controlled LMP can therefore run together. The
has been established. The network PBB-TE. While standardization is far two IETF drafts only specify how
center intends to evaluate the feasibility from complete, two GELS drafts have GMPLS is used to build point-to-point
of PBB-TE in the campus network by been released[7]. These drafts involve PBB-TE paths. Improvements are
comparing the PBB-TE convergence GELS architecture and technical required in the specifications of
network with the existing IP specifications. GELS uses as many of point-to-multipoint path establishment
convergence network. A study of the the original GMPL components as and control-plane-based protection
PBB-TE control plane will also be possible, and makes necessary recovery. 802.1aq Provider Link State
conducted on equipment connected to extensions. Bridging (PLSB) is also a solution for
MPLS switches. (1) Addressing Mode PBB-TE control[9].
The node on the GELS control plane
still uses the IP address ID, and control 2.2 Simulation Platform of GMPLS
2 GMPLS Controlled plane messages are exchanged on the Controlled PBB-TE
PBB-TE IP layer. GELS supports both labeled Researchers are likely to use a virtual
Although PBB-TE control plane and unlabeled ports. PBB-TE testing platform rather than a
standards are still being studied, (2) Signaling Protocol platform built with real PBB-TE
Special Topic Features of PBB-TE Architecture and GMPLS Control Technology
metropolitan area networks: virtual bridged local

●
IP Layer ●
● area networks, amendment 7: provider backbone
●
● ●
● bridge traffic engineering [S]. 2009.
● ● [5] IEEE 802.1Qaw. IEEE standard for local and
●
● metropolitan area networks: virtual bridged local
●
●
●
● area networks, amendment 9: management of data
●
● driven and data dependent connectivity faults [S].
●
2009.
● ●
[6] Deutsche telecom flirts with PBT [EB/OL].
●
[2007-09-20]. http://www.lightreading.com/
document.asp?doc_id=134344.
Optical ■ ■ [7] Generalized Multi-Protocol Label Switching
Layer (GMPLS) Ethernet label switching architecture and
■
■ framework [R].
draft-ietf-ccamp-gmpls-ethernet-arch-09. 2010.
◀Figure 4.
■
■
■ [8] Ethernet traffic parameters [R].
■ Screenshots of the Platform for draft-ietf-ccamp-ethernet-traffic-parameters-10.
■ 2010.
■ Large-Scale Optical Network [9] ALLAN D, ASHWOOD-SMITH P, BRAGG N, et al.
Validation. Provider link state bridging [J]. IEEE
Communications Magazine, 2008, 46(9): 110-117.
equipment because a virtual platform is function on the data plane. A node on [10] SOBIESKI J. DRAGON: Dynamic resource
allocation via GMPLS optical networks [C]//MCNC
more flexible and supports more nodes. the platform is only an object in the Optical Control Planes Workshop, Apr 23, 2004,
Such virtual platforms are of two types: computer’ s memory; signaling Chicago, IL, USA.
(1) Emulation interaction between nodes as well as
An emulation platform is represented routing message updates are Biographies
by Finite State Machine (FSM) implemented by communications Wei Jianwen
emulation software such as Network between objects, without the need for
Simulator 2 (NS2). It supports a large actual network cards. Both the Wei Jianwen is studying for a
number of nodes and has good signaling and routing messages are master’s degree at the
Electronic Engineering
scalability; however, it lacks details of recorded in logs for offline reading. The Department of Shanghai Jiao
signaling interaction, and fidelity of the platform is intended to successfully Tong University. His research
direction is multi-layer
control and data planes is not good demonstrate cross-layer path multi-domain networks,
enough. establishment where tens of nodes are including TE and segment
protection in multi-domain
(2) Simulation involved. GMPLS extensions are to be
networks, and survivability of
The simulation platform is implemented on the platform for multi-layer networks.
represented by the DRAGON support of PBB-TE protection
program[10], and uses computers to switching. Xie Rui
replace PBB-TE bridges. The complete
GMPLS protocol stack is run in the Xie Rui is a senior engineer at
computers, and data frames are sent 3 Conclusion the Network&Information Center
of Shanghai Jiao Tong
through the network cards. Therefore, PBB-TE is a PTN technology with University. He is responsible for
the PBB-TE control and data planes layered architecture, improved OAM, the operation and maintenance
management of both the
can be realistically simulated. and QoS guarantee. As a Shanghai Education and
Scalability is limited because one convergence-layer solution, PBB-TE is Research Network and the
Shanghai Jiao Tong University
PBB-TE bridge requires one computer more cost effective than MPLS. website. His research interests
for simulation. PBB-TE and GMPLS standards are still include key technologies and
This paper focuses on signaling in progress, and more telecom network applications of next-generation Internet, network
management, network measurement, and network
interworking and cross-layer characteristics will be introduced into security. He has published 10 academic papers.
optimization of GMPLS, and so a the PBB-TE system. With improved
compromise is made between standards, PBB-TE is expected to Jin Yaohui
emulation and simulation. As shown in become an optimal solution for
Figure 4, the large-scale optical next-generation metro PTNs. Jin Yaohui is professor and
network validation platform simulates a doctoral advisor at the State Key
References Lab of Advanced Optical
2-layer network, with PBB-TE on the [1] IEEE 802.1Q. IEEE standards for local and Communication Systems and
upper layer and SONET/SDH on the metropolitan area networks: virtual bridged local Networks, Shanghai Jiao Tong
area networks [S]. 2003. University. He is also the deputy
bottom layer. Resource Reservation [2] IEEE 802.1ad. IEEE standards for local and director of the
Protocol-Traffic Engineering metropolitan area networks: virtual bridged local Network&Information Center of
(RSVP-TE) and OSPF-TE are area networks, amendment 4: provider bridge [S]. the University. His research
2005. interests include future Internet
implemented completely on the [3] IEEE 802.1ah. IEEE standards for local and design, large-scale network
platform, as are the GMPLS extensions metropolitan area networks: virtual bridged local measurement and optimized control, and Photonic
area networks, amendment 6: provider backbone Network-on-Chip (PNoC). He has published 100
to PBB-TE and SONET. The platform bridge [S]. 2005. academic papers, of which 30 are indexed by SCI.
does not implement the forwarding [4] IEEE 802.1Qay. IEEE standards for local and
Ring Protection and Survivability Mechanisms for Packet Transport Networks Special Topic
Lu Yueming
Ring Protection and Survivability

Mechanisms for Packet
Transport Networks
Lu Yueming
(Key Laboratory of Information Photonics and Optical Communications of Ministry of Education, Beijing University of Posts and Telecommunications,
Beijing 100876, P. R. China)
Abstract:
Packet Transport Networks (PTNs) must resolve issues in their protection and recovery mechanisms. These issues include the detection
performance of Operation, Administration and Maintenance (OAM), network resource optimization, resource allocation deadlock, and
resource deployment blocking. Traditional protection and recovery mechanisms cannot meet the requirements of PTN. In order to
improve the protection and recovery performance, network resource use, probability of service recovery, and to decrease the
probability of service blocking, this paper introduces overlapped segment shared protection, Pre-configured Multi-Cycle (P-mcycle),
conflict-free algorithm, and delay restoration algorithm. Only with appropriate protection and restoration mechanisms is it possible to
achieve smooth evolution from TDM networks to integrated packet-based bearer networks and all-IP services.
T
he development of metro flexible dynamic control are added for protection) of a PTN network, and thus
transport networking has been adaptation to data services. These are affects services. The survivability of
driven by access and basic but important technologies for PTN involves blocking during Label
transmission demands of network convergence. In terms of Switching Path (LSP) setup, resource
broadband data services (e.g. 3-play), transport mode, PTN not only provides utility, and coordination for resource
Ethernet Private Line (EPL) enterprise packet data services and traditional allocation[4]. Unlike SDH networks, PTN
services, Layer 2 Virtual Private Time Division Multiplexing (TDM) protection mechanisms are still being
Network (L2VPN) services, and services, but also offers Asynchronous studied and protection performance is
common broadband services. As Transfer Mode (ATM), Inverse far from adequate.
service and bearer networks move Multiplexing for ATM (IMA)[2], and Protection and restoration strategies
towards IP, basic transport networks Multi-Level Pre-Emptive Priority of PTN play an important role in the
are evolving into Packet Transport (MLPPP) to meet transmission network’ s service performance,
Networks (PTNs)[1]. demands of 3G mobile systems. PTN resource utility, and survivability.
With good scalability, powerful OAM, devices adopt several encapsulation Because PTN imposes high
and fast protection switching, PTN and adaptation technologies, requirements on survivability, research
inherits some of the characteristics of integrating various TDM and data on its protection and restoration
traditional Synchronous Digital services on a unified packet transport strategies is important. Only with
Hierarchy (SDH) transport networking. plane. PTN is thus a unified, adequate protection and restoration
New features such as packet switching, packet-switched multi-service mechanisms is it possible to evolve
statistical multiplexing, transmission platform[3]. smoothly from TDM-based transport to
connection-oriented label switching, However, integration of various packet transport, to achieve network
Quality of Service (QoS) guarantee, and demands and diversified application convergence and to deliver all-IP
scenarios creates some critical services.
problems for survivability. This issue
This work was funded by the National High directly impacts both the networking
( program) under Grant mode (ring, mesh, or star) and the 1 PTN Research Status
No. 2007AA01Z252. service setup mode (1+1 or 1:1 path There are currently two main
Special Topic Ring Protection and Survivability Mechanisms for Packet Transport Networks
Lu Yueming
technologies for implementing PTN: carrier-class protection switching of carrier-class PTNs can occur.
Transport Multiprotocol Label Switching strategies. Key technologies being Challenges for PTN include survivability
(T-MPLS)[5]/Transport Multiprotocol considered include fault detection and evaluation and optimization, distributed
Label Switching Traffic Engineering fault notification. These technologies deadlock and resource contention,
(MPLS-TE)[6] , and Provider Backbone aim to achieve an overall switching time PTN-oriented protection and
Bridge (PBB)/Provider Backbone of less than 50 ms, supporting MPLS restoration blocking, as well as fault
Transport (PBT)[7]. Research into FRR protection (by enabling local detection of OAM.
survivability strategies for PTNs with protection of important links and (1) Survivability Evaluation and
these 2 technologies has attracted network elements in the network Optimization
much attention. Internationally, topology), and supporting Link Among these challenges,
institutes and organizations such as Aggregation (LAG) protection of survivability evaluation and optimization
Alcatel[8], Nortel[7], Cisco[9], Sycamore[10], the links. is particularly troublesome. First,
European Union Seventh Framework In 2008, the China Communications determining reasonable criteria for
Programme (FP7)[11], the University of Standards Association (CCSA) evaluating survivability is difficult.
Texas[12], Stanford University[13] and Bell established a forum specifically Second, there are few good
Labs[14] are carrying out research. In focused on transport devices based on optimization methods and technologies
China, telecom equipment a unified switching plane. In 2010, the for PTN survivability. Perhaps due to
manufacturers such as ZTE, Huawei, forum has turned to T-MPLS protection limitations in human cognitive ability
and Fiberhome are studying mechanisms. when compared with computing
survivability strategies, while research Overall, current research into PTN capability, many optimization issues
institutes at Tsinghua University, Peking protection mechanisms is focused on seem particularly difficult to solve.
University, Shanghai Jiaotong improving T-MPLS 1+1 and 1:1 linear People simply do not know what the
University, and Beijing University of protection—as it no longer meets the best network is. Moreover, limited
Posts and Telecommunications are demands of various services. PTN computing capability makes the
initiating research into protection and requires multi-level multi-layer method of exhaustion infeasible. Only
restoration mechanisms from the protection and restoration. To achieve algorithms such as genetic and greedy
perspective of resource management, this, path-level and link-level algorithms can be used to find a better
distributed parallel resource allocation, protection as well as quick routing or solution to survivability optimization in
and contention. rerouting of services by Generalized distributed networks.
The International Telecommunication MPLS (GMPLS) intelligent software is (2) Distributed Deadlock and
Union-Telecommunication necessary. Resource Contention
Standardization Sector (ITU-T) and the In fact, precursors to packet network To a greater or lesser extent, PTN
Internet Engineering Task Force (IETF) protection mechanisms already exist. may encounter distributed deadlock
have brought forward their own W. Grover, for example, introduced the and resource contention problems
solutions. ITU-T defines T-MPLS, and concept of P-cycle[17] (designed for IP when establishing a fast LSP,
supports 1+1 and 1:1 linear protection networks). Before a fault occurs, a set preempting a fast LSP in 1:N protection,
(G8131) as well as wrapping and of P-cycles are pre-configured and or reselecting a quick LSP. At present,
steering ring protection (G8132)[15]. IETF routed through all nodes and links to be the ostrich algorithm is commonly used
prefers the Fast ReRoute (FRR)[16] protected, and resources are reserved to solve these problems. This algorithm
technique of Multiprotocol Label for them. When a fault occurs, switching meets the minimum requirements of IP
Switching (MPLS) to implement 1:N is performed according to the networks, but cannot realize fast
linear and ring protection. In 2008, IETF pre-designed ring. In this way, ring protection and restoration. Distributed
and ITU-T established a Joint Work protection is achieved. In a public deadlock is an awkward problem
Team (JWT), and currently, experts patent, Robert Sultan proposes a server affecting computer operating systems
within this team are studying ring trail failure message notification and networks, and it has yet to be
protection requirements of MPLS mechanism. This increases the solved. For computer operating
Transport Profile (MPLS-TP). Limited probability of service recovery in cases systems, Dijkstra proposes banker’ s
by its global labels, PBB Traffic of network congestion[18]. Raymond Xie algorithm, but its effectiveness has
Engineering (PBB-TE) supports 1:1 of Sycamore suggests an intelligent been proven poor. The deadlock
linear protection, but does not support protection and restoration mechanism problem receives little attention in the
subnetwork protection and in EtherOptics-based PTN[19-20]. research of MPLS; however, in GMPLS
connection-based ring protection. research, Zafar Ali et al have begun to
In China, PTN protection and think deeply about the problem and
restoration technologies are still in the
2 Challenges for PTN have addressed it to the IETF.
early stages of research and Survivability Currently, studies into deadlock are
discussion, with network device Survivability is an important issue that underway.
manufacturers focused on must be addressed before deployment (3) Blocking in Deploying
Lu Yueming
PTN-Oriented Protection and

2
Restoration 2 3′
Some protection mechanisms similar 0.6 0.6

to SDH have already been adopted in
PTN. However, in distributed routing 1 5 1 3
0.6
3′ 5
environments rigid protection paths are
set up, and blocking with these
mechanisms is very high. To solve this 3 4 4
problem, lenient protection and
restoration mechanisms must be (a) Non-Overlapped Segment Protection (b) Overlapped Segment Protection
worked out to complement traditional
rigid ones, and to meet the specific ▲Figure 1. Non-overlapped and overlapped segment protection.
requirements of PTN.
(4) Fault Detection of OAM provides multiple overlapped protection (2) Pre-configured Multi-Cycle
OAM detects Peer-to-Peer (P2P) segments in the entire working path, (P-mcycle) Scheme
connectivity status using connectivity enabling reasonable and effective To resolve uncertain restoration of a
detection messages which interact selection of these overlapped fault point in mesh networks, a new
periodically. Along with other methods, protection segments. Compared with restoration scheme, P-mcycle, is
it detects packet loss rate, delay, and traditional protection algorithms, the introduced. This scheme uses Integer
jitter, and thus achieves loop-back, overlapped segment shared protection Linear Programming (ILP) optimization
alarm suppression, and alarm mechanism enhances network algorithm (for P-mcycle generation), as
feedback. Moreover, OAM uses connection reliability, and improves the well as Sub-cycle secondary routing
Automatic Protection Switching (APS) utility of network resources by algorithm. P-mcycle ensures a valid
protocol to protect the two End-to-End reasonably sharing resources among fault restoration time, greatly enhances
(E2E) paths. Whatever the situation, overlapped protection segments. fault protection success rate, and
SDH-like protection mechanisms must This mechanism protects the entire enchances the efficiency of protection
provide carrier-class protection, and end-to-end working path using resource reservation. Thus, it improves
this implies a switching time of no more multiple backup protection segments. effectiveness and reliability of
than 50 ms. The time for the OAM Compared with path protection, its restoration in PTN.
engine to complete path fault detection limited restoration scope is a major P-mcycle generation algorithm is
should not be too long. Specifically, to advantage. When a link or node fails to based on ILP. It introduces a P ring for
ensure an overall switching time of less work, only its corresponding protection fault-independent path protection.
than 50 ms, fault detection and switch segment is activated. As a result, However, the algorithm involves
triggering must be done 25 ms after a restoration time is shortened. The complex computation which often takes
fault occurs. Therefore, OAM fault mechanism does not provide a a long time. For a small network,
detection performance is very important. protection segment for each link in the P-mcycle generation takes 1 hour; and
working path, so it achieves higher for a large network, 2 days. In contrast,
3 Ring Protection and resource utility than link protection. ILP optimization algorithm leaves
Instead, it provides a backup variables and constraints such as
Survivability Mechanisms overlapped protection segment for dynamic service distribution and
for PTN each segment in the working path, service symmetry to secondary routing
Most core networks of PTN adopt a allowing some working links to fall algorithm. A pure ILP model is only
mesh structure, so ring protection under the protection of several backup used for sourcing optimal P-ring
differs to that of SDH. Here, research segments. This enables flexible network combinations. In this way, the
carried out by BUPT concerning ring resource allocation and increases the computation involved in P-mcycle
protection and survivability survivability of the network. Traditional generation decreases considerably,
mechanisms of PTN is introduced. non-overlapped segment protection and the generation task becomes static
(1) Overlapped Segment Shared mechanisms cannot protect nodes planning of network topology.
Protection Mechanism between segments. Overlapped With the number of variables and
After intensive analysis of constraint segment protection mechanism constraints decreased, and
mechanism and overlapped segment resolves this problem. With neighboring computational complexity dramatically
protection mechanism, the 2 segments overlapped, a faulty node reduced, an effective pre-selection
mechanisms were integrated, and between segments will not lead to method can be used to narrow down
overlapped segment shared protection failure of both working and protection the number of possible solutions to an
algorithm[21] was used to dynamically segments. Overlapped and ILP problem. In a given network
adjust the working and protection paths non-overlapped segment protection is topology, enhanced Depth-First
based on link weights. This algorithm illustrated in Figure 1. Search (DFS) algorithm is used first to
Special Topic Ring Protection and Survivability Mechanisms for Packet Transport Networks
Lu Yueming
B the path message reaches a node, and

Patherr: Path Error
Resv: Reservation the new value is stored at the node.
i 1=4 Resverr: Reservation Error When the path message arrives at the
Re
sv
1 destination node, the value of the
counter corresponds to the number of
Pa
Resv 1_i =2 ＜ i 2=4

th
1
links needed to establish a path from

D i 1=0
Resv1 Resv1 Resv1 the source node to the destination
A i 1=3 i 1=2 i 1=1 E
Path1 Path1 Path1 node. As shown in Figure 2, to establish
Path2 Path2 Path2 Path2 Path2 a path from Node A to Node E, 5 links
i 2=0 i 2=1 i 2=2 i 2=3 Resv2 i 2=4 Resv2 i 2=5 must be passed; from Node B to Node
Pather2 Pather2 Pather2
C E, 4 links must be passed.
Pather2 Pather2
Using conflict-free algorithm for
Resv 2_i =4 ＞ i 2=2 establishing long-distance paths is
unlikely to be successful. Suppose
▲Figure 2. Conflict-free algorithm. there are 2 path establishment
requests: Request A involves 7 links,
search all simple rings. Then, the resources. and Request B involves 3 links. If the
searched rings are checked one by one (3) Conflict-Free Algorithm conflict link is 2 links away from the
to determine whether their links satisfy In establishing paths, 2 nodes source node of Request A, and one link
the constraints. M rings with the largest request the same resources (port or away from the resource node of
AE values are selected to form a wavelength λ) at both ends of a link and Request B, Request B will succeed in
sub-optimal solution space. Finally, ILP send Reservation (Resv) messages. establishing a path but Request A will
optimization algorithm is used to select When the messages reach the other fail. As a result, the 5 paths that have
a group for P-mcycle. P-mcycle end of the link, the requested resources been established by Request A before
obtained in this way can protect all may already be occupied. In traditional the conflict must be released, resulting
working wavelengths, and ensure the solutions, the node receiving the Resv in resource waste.
preset maximum reserved wavelength message treats this as a path error, and (4) Delay Recovery Algorithm
is not exceeded. Service is 100% returns a Reservation Error (Resverr) Delay recovery algorithm is used for
restored with the least resources. message. If resources at both ends are handling network faults, adopting
Secondary routing algorithm occupied, the nodes at both ends sequenced delay during network fault
simplifies pre-configuration determine that their paths have not recovery.
computation in traditional P-cycle, and been established. As a result, the After a fault occurs, the algorithm
optimizes protection path selection resources used for path establishment allows the source node of the service to
during switching. With P-mcycle, fault have been wasted. send a delayed Notify message. As
restoration time satisfies the rigid Conflict-free algorithm avoids such a shown in Figure 3, a time interval t is
survivability requirements of PTN. problem, as shown in Figure 2. This inserted between two Notify messages.
Compared with traditional P-cycle, method works on the principle of With this method, the probability of
P-mcycle has a higher protection adopting a priority comparison strategy resource conflict is reduced in the
success rate, better resource utility, for the Resv messages sent at the same re-routing and signaling process for
and better efficiency. Simulation results time, enabling one path to be fault recovery. When services in two
show that optimization is more established[22]. Suppose conflict takes opposite directions are recovered,
pronounced in the Mesh network where place between Node C and Node D, signaling messages are sent in a
the average node density is low. and the resource allocated by Node C certain sequence and no resource
P-mcycle research will continue to for Path 1 is the same as that allocated conflict occurs.
focus on the following aspects: by Node D for Path 2. In traditional
achieving P-mcycle optimization solutions, Resv 1 message sent by
based on node fault restoration and Node C arrives at Node D, the 4 Conclusion
multi-link/multi-node restoration; requested resource has been PTN is a future direction in telecom
enabling reasonable distribution of occupied, and Node D treats it as a network development. However, ring
P-mcycle generation resources by path error. Similarly, Node C registers a protection and survivability directly
taking into account dynamic service path error when it receives Resv 2 impact its QoS. PTN must resolve
distribution changes during candidate message from Node D. Therefore, issues in fault detection, network
ring selection; and studying traffic neither path can be established. deployment, protection and restoration.
engineering involved in secondary In conflict-release algorithm, a Existing protection and restoration
routing based on the direction of counter i is carried along with a path methods cannot meet the requirements
services in order to improve service message. The counter is set to 0 at the of PTN and new mechanisms must be
protection success with fewer network source node. A "1" is added whenever worked out. These challenges will
Lu Yueming
mesh-like capacity for self-planning network

A B C D restoration Communications [C]//Proceedings of
International Conference on Communications (ICC'
Notify1 98): Vol 1, Jun 7-11, 1998, Atlanta, GA, USA.
Notify2 t Piscataway, NJ, USA: IEEE, 1998: 537-543.
[18] Packet transport: it’s time to clear up the
Path1
Path1 confusion [J]. Fibre Systems Europe, 2008(3): 21.
Path1 [19] SULTAN R, YONG L. Dunbar l. Network availability
Path2 enhancement technique for packet transport
Path2 Resv1 networks: US, 370235 [P]. 2008-11-13.
Path2 Resv1 [20] XIE R.“EtherOptics”optimizes intelligent IP/
Resv1 Ethernet transport [J]. Lightwave, 2008, 25(5).
Resv2 [21] 孙翔, 陆月明, 纪越峰. 基于 SRLG 约束和资源共享的
Resv2 交迭段保护机制研究 [J]. 光子学报, 2009, 38(6):
1386-1391.
Resv2
SUN Xiang, LU Yueming, JI Yuefeng. Overlapped
Resv: Reservation
◀Figure 3. segment shared protection scheme based on
Delay recovery algorithm. SRLG constraints [J]. ACTA PHOTONICA SINICA,
2009, 38(6): 1386-1391.
[22] 陆月明, 赵同, 弓伟丽, 等. 光网络实时连接恢复中资
eventually be solved in the future. msg09503.html. 源分配死锁解除方法 [J]. 通信学报, 2008, 29(12):
[9] Cisco. Multipoint TMPLS [EB/OL]. [2007-01-25]. 91-95.
http://www.ietf.org/ mail-archive/web/pwe3/current/ LU Yueming, ZHAO Tong, GONG Weili, et al.
References msg08706.html. Elimination methods of resource allocation
[1] NEC. Packet transport network tech makes NGNs [10] 2nd VIOLA Workshop, Mar 22-23, 2006, Bonn, deadlock for optical networking real-time
reliable [EB/OL]. [2007-04-26]. http://www.eetindia. Germany [C]//e-Photon/ONe+/IST Phosphorous connection restoration [J]. Journal on
co.in/ART_8800462498_1800006_NT_e79940b5. Meeting, Sep 16, 2007, Berlin, Germany. Communications, 2008, 29(12): 91-95.
htm. [11] YOO S J B. Network interoperability—the role of
[2] LEE D S, SENGUPTA B. Queueing analysis of a optical technologies [C]// e-Photon/ONe+/IST
threshold based priority scheme for ATM networks Phosphorous Meeting, Sep 16, 2007, Berlin,
[J]. IEEE/ACM Transactions on Networking, 2003, 1 Germany.
(6): 709-717. [12] Fall 2008 graduate research day [R]. Dallas, TX,
[3] VAINSHTEIN S, FROST T. Emulation of TDM circuits USA: University of Texas. 2008. Biography
over MPLS using raw encapsulation implementation [13] Ixia to provide IP testing horsepower in the largest Lu Yueming
agreement [R]. MFA Technical Committee, MFA public carrier Ethernet interoperability test [EB/
8.0.0. 2004. OL]. [2008-09-24]. http://www.reuters.com/article/
[4] ITU-T Study Group- Question 9/15.Transport pressRelease/idUS181350+22-Sep-2008+ Lu Yueming received his PhD
equipment and network protection/restoration [R]. BW20080922. degree from Xi’ an Jiaotong
2009. [14] Alcatel-Lucent Bell Labs. Alcatel-Lucent Bell University, and is currently a
[5] ITU-T Recommendation G.8110.1/Y.1370.1-2006. Labs and NTT Photonics Labs sign collaborative professor at Beijing University
Architecture of transport MPLS (T-MPLS) layer agreement related to optical packet switch of Posts and
network [S]. 2006: 8-15. technology [EB/OL]. [2008-05-05]. http://goliath. Telecommunications. He is
[6] IETF RFC 3812. Multiprotocol Label Switching ecnext.com/coms2/gi_0199-7956799/ also a member of the China
(MPLS) Traffic Engineering (TE) [S]. 2004. Alcatel-Lucent-Bell-Labs-and.html. Computer Federation. He has
http://www2.nortel.com/go/solution_content.jsp? [15] LEE, B J, BACQUE J B. Protection system and been engaged in researching
segId=0&catId=0&parId=0&prod_id=55120. method for Resilient Packet Ring (RPR) computer networks, packet
[7] Nortel. Provider backbone bridges (PBB) and interconnection: US, 7274656 [P]. 2007-09-25. transport networks, and
provider backbone transport (PBT) [EB/OL]. [16] MPLS FRR protection overview [EB/OL]. intelligent optical networks for years. He has
[2006-02-15]. [2009-04-01]. http://www.juniper.net/ techpubs/ presided over and participated in 5 nationally
[8] Alcatel. PWE3 response to ITU-T SG15Q9 Liaison software/nog/nog-mpls-frr/html/frr-overview.html. funded projects, has been granted 11 patents, and
on TMPLS ring protection [EB/OL]. [2008-02-09]. [17] GROVER W D, STAMATELAKIS D. Cycle-oriented has published over 70 papers.
http://www.ietf.org/mail-archive/web/pwe3/current/ distributed preconfiguration: ring-like speed with
Roundup
•• ••• • •• ••• ••• ••• • •• ••• • •• ••• ••• ••• • •• ••• • •• ••• ••• ••• • •• •••
•• ••• • •• ••• ••• ••• • •• ••• • •• ••• ••• ••• • •• ••• • •• ••• ••• ••• • •• •••
ZTE Tops the Global CDMA Market with a 30% Market Share
•• ••• • •• ••• ••• ••• • •• ••• • •• ••• ••• ••• • •• ••• • •• ••• ••• ••• • •• •••
•• ••• • •• ••• ••• ••• • •• ••• • •• ••• ••• ••• • •• ••• • •• ••• ••• ••• • •• •••
•• ••• • •• ••• ••• ••• • •• ••• • •• ••• ••• ••• • •• ••• • •• ••• ••• ••• • •• •••
•• ••• • •• ••• ••• ••• • •• ••• • •• ••• ••• ••• • •• ••• • •• ••• ••• ••• • •• •••
IDC’s 2010 Global CDMA Market of its CDMA products. It has wireless capacity of 250 million lines.
and Prospects research note said that consolidated its lead in the key A total of four trial and commercial
by Q1 2010, ZTE had shipped an emerging markets of China, Indonesia EV-DO Rev.B networks were
accumulated total of over 250,000 and India, and made breakthroughs in deployed worldwide including the
base stations, allowing the Chinese North America. In the Chinese market world's first commercial EV-DO Rev.
company to claim the No. 1 spot in the in 2010, ZTE has undertaken over 80% B network in Indonesia. Meanwhile,
CDMA market with a global share of of the work to migrate Nortel and ZTE has also cooperated with seven
30.3%. Motorola CDMA equipment. top-tier operators worldwide to
ZTE is the first vendor in the industry By the first quarter of 2010, ZTE’s deploy commercial LTE networks and
to launch the CDMA/LTE dual mode CDMA products were used by more almost 50 trial networks in Europe,
system. ZTE has also steadily than 120 operators in over 70 America, Asia Pacific and the Middle
enhanced its overall competitiveness countries with an accumulated East. (ZTE Corporation)
Special Topic Service Adaptation and Label Forwarding Mechanism for MPLS-TP
Zhang Yongjun，Zhang Zhihui，Gu Wanyi
Service Adaptation and Label

Forwarding Mechanism for
MPLS-TP
Zhang Yongjun, Zhang Zhihui, Gu Wanyi
(Key Laboratory of Information Photonics and Optical Communications of Ministry of Education, Beijing University of Posts and Telecommunications,
Abstract:
Most services borne by transport networks have been transitioned away from Time Division Multiplexing (TDM) towards IP-based. New
technology is therefore required to effectively transport packet services. Multiprotocol Label Switching Transport Profile (MPLS-TP) is
such a technology which is connection-oriented, capable of multiservice adaptation, and has a flexible label forwarding mechanism.
This paper analyzes the characteristics of MPLS-TP, and introduces its latest standardization development. Key issues including data
forwarding plane, service adaptation, and label forwarding mechanism are discussed, as well as service implementation of Virtual
Private Network (VPN) based on dual-label transport. MPLS-TP applications are summarized in conclusion.
S
ervice networks are undergoing There are 2 solutions available for room for improvement.
transformation due to the PTN implementation: enhanced
shrinkage of traditional telecom Ethernet technologies, and transport
services and the trend towards technology combined with
1 MPLS-TP Features and
IP-based services. The future market Multiprotocol Label Switching (MPLS). Standardization
will require packet transport technology Provider Backbone Bridge Traffic
that transmits packet services Engineering (PBB-TE)—or Provider 1.1 MPLS-TP Standardization
effectively, but also provides Backbone Transport (PBT)—is Since 2005, Study Group 15 of the
carrier-class Operation, Administration representative of the former; MPLS International Telecommunication Union
and Maintenance (OAM), as well as Transport Profile (MPLS-TP)—or Telecommunication Standardization
protection. Transport devices must Transport MPLS (T-MPLS)—is Sector (ITU-T SG15) has focused on
change from being adaptable to representative of the latter. Compared using MPLS technology to define
multiservice interfaces into being with other technologies, these better packet transport layer’ s service
adaptable to multiservice kernel. In serve the goals of PTN, and have come function structure (T-MPLS
other words, IP-based services require into mainstream use. Both are technology). In 2007, ITU-T made
packet transport networks; but at the connection-oriented, have recommendations on aspects such as
same time, traditional services such as performance and reliability similar to system framework, interface and device
voice should be delivered normally. To Synchronous Digital Hierarchy (SDH), specifications, OAM, protection
achieve this, packet technologies and and offer standard, switching mechanism, and service
transport technologies must be connection-oriented tunnels. Their signal adaptation. These
integrated. And in this context, Packet main differences lie in the way they recommendations have received
Transport Network (PTN) is introduced[1]. implement data forwarding, protection support from the Internet Engineering
and OAM. In the process of Task Force (IETF). At present, ITU-T is
standardization, T-MPLS has outpaced working on detailed definitions of all
This work was funded by the National High PBT. T-MPLS standards have matured, layer functions, addition of adaptive
( Program) under Grant and meet requirements for client signals, and service
No. 2008AA01A328. commercialization, although there is still interconnection and synchronization.
Service Adaptation and Label Forwarding Mechanism for MPLS-TP Special Topic
IETF is preparing T-MPLS Request for connectionless features are excluded the payload convergence sub-layer.
Comments (RFC), and defining new but protection switching and OAM The functions of the 3 sub-layers are
labels for T-MPLS services. In 2008, functions recommended by the ITU-T described below.
ITU-T and IETF established a Joint are present. All these facilitate (1) The payload convergence
Work Team (JWT) to promote the carrier-class services. MPLS-TP also sub-layer is primarily responsible for
merging of T-MPLS and MPLS. IETF retains MPLS features that are helpful encapsulating the payload in VC
expanded existing MPLS technology for data service transport, discards the Protocol Data Units (VC-PDUs). It
into MPLS-TP to better support complicated MPLS control protocol carries the additional information
transport requirements specified by cluster defined by IETF, simplifies the needed to replay the native data units
ITU-T. In the future, JWT will develop data plane, and removes unnecessary at the Customer Edge (CE) boundary.
MPLS-TP standards and ensure forwarding processing. As a result, it is When a bitstream is sent to MPLS-TP,
consistency between T-MPLS and suited to situations where TDM-based one part is peeled off the native service
MPLS-TP standards[2]. services are being transitioned into processing module. For example, in
IP-based ones. structured SDH, fragmentation and
1.2 MPLS-TP Technical Features circuit overheads may be peeled off.
MPLS-TP has a clear framework and (2) The sequencing sub-layer
complete solutions for implementing its
2 MPLS-TP Service performs three functions: frame
key techniques. It simplifies and Adaptation Technologies ordering, frame duplication detection,
innovates on MPLS/Pseudo Wire (PW) The unified transport/aggregation and frame loss detection. Some
technology, introducing concepts such platform is a breakthrough in service services must be delivered in order,
as PTN layer, OAM and linear networking technologies. Service while other services need not. The
protection. Thus, it meets the demands providers need not invest in each choice of strategy for handling incorrect
of PTN[3]. As a connection-oriented network layer to transport mixed frame sequence, or detected frame
packet transport technology, MPLS-TP services. In MPLS-TP, the transport duplication or loss depends on the
consists of a data plane, management plane is responsible for packet service type. Client signals such as
plane, and control plane. An transport of client data, and also SDH and Frame Relay (FR) may have
end-to-end connection-oriented adapting to and forwarding client clock recovery and timed packet
packet transport tunnel is also signals. For different client signals, delivery timing requirements.
established. This tunnel, set up with the different signal adaptation and (3) The timing sub-layer performs
network management system or forwarding methods are employed. two functions: clock recovery and timed
intelligent control plane, is packet data, cell data and TDM data delivery. Clock recovery involves
characterized by good operability, and are quite different in length, format, and extracting clock information from the
maintenance and protection recovery. multiplexing mode. Therefore, different delivered bit stream, and recovering
MPLS-TP maps client signals into aggregation, fragmentation, the clock with a phase-locked
MPLS frames and forwards them using encapsulation, sequencing, timing and mechanism. Timed delivery involves
mechanisms such as label switching or multiplexing/demultiplexing methods delivering non-contiguous VC-PDUs
label stack. Basic functions of the are used in their adaptive transport[7]. to CEs with a constant phase relation.
transport layer—including connection IP services can be directly mapped
and performance monitoring, 2.1 Signal Adaptation into T-LSP or indirectly mapped by
survivability and management and Client signals can be directly mapped means of dual labels. In dual-label
control planes—are included[4-5]. into T-LSP, or indirectly mapped based encapsulation, a node does not need
MPLS-TP adopts a dual-label on CII. In the dual-label architecture, L3 forwarding capability. Hence, in
transport mode[6]. When it provides all types of service signals can be networks where nodes do not have L3
packet data transport for the client encapsulated in a dual-label structure. forwarding capability, dual-label
layer, it assigns 2 labels to client data: The encapsulation layer structures the encapsulation is advantageous. With
Common Interworking Indicators (CII) specified payload signals for non-IP service adaptation, VC-based
and Transport Label Switched Path transmission over the Virtual Circuit indirect mapping is adopted. As
(T-LSP). The CII label associates the (VC). It consists of 3 sub-layers: multiple VC multiplexing is transferred
clients at both ends, and is used by payload convergence, timing, and in a Label Switched Path (LSP), the
terminals to distinguish client data, sequencing. The payload convergence service granularity can be less than
while T-LSP label is used for exchange sub-layer is tailored to the specific 2 Mbit/s. Moreover, with CII and label
and forwarding client data in the payload type. It can group a number of stack added into the services, address
MPLS-TP packet data tunnel. payload types into a generic class, and space limitation—which occurs in some
In short, MPLS-TP is a subset and then provide a single convergence technologies—can be overcome.
abbreviated version of MPLS. To sub-layer type that is common to the
support a connection-oriented group. The timing and sequencing 2.2 Service Encapsulation
end-to-end OAM model, many sub-layers provide generic services to The generic encapsulation format of
monitoring.
Payload (1) Pre-processing refers to the
Payload Convergence operations on client data before
switching. These operations include
Timing
data and address conversion, and
Sequencing identification of client data type.
CII Multiplexing Pre-processing simplifies the design of
the processing steps that follow.
MPLS-TP Convergence
(2) The convergence module is
MPLS-TP mainly responsible for sorting and
Payload
Data Link Layer converging packets based on type,
Flags Fragment Length Sequence Number and importance of client data signals or
Physical Layer
CII Label signaling signals. It dispatches packets
to different transport channels, so that
T-LSP Label
different signal types receive different
CII: Common Interworking Indicators T-LSP: Transport Label Switched Path classes of Quality of Service (QoS).
MPLS-TP: Multiprotocol Label Switching Transport Profile
(3) The encapsulation module adapts
▲Figure 1. Generic encapsulation format of services. the signals before they are multiplexed
with T-LSP and forwarded. The
services is illustrated in Figure 1. The transport mode in which L3 IP is used to workings of the module are closely
payload containing L2 header or L1 transmit L2 Ethernet data. The 2 labels related to the client signal type to be
overhead can be IP packet, Ethernet are CII and T-LSP. To support encapsulated. For example, different
packet, ATM cell, FR cell, or SDH MPLS-TP networking, T-LSP labels encapsulation methods are used for
payload. Data information and control can be infinitely embedded. Therefore, packets, cells, and Time Division (TD)
words are used for payload there may be several T-LSP labels. The signals. Encapsulation is the process of
convergence. With CII label built in, the CII label can be a client signal label; for attaching VC and T-LSP labels onto the
VC type in T-LSP can be determined; example, in Figure 2, the CII label refers packets, and inserting OAM information
with T-LSP label built in, MPLS-TP LSP to VC labels. The into the packets. Depending on the
can be determined. Control words often multiplexing/demultiplexing module client signal type, 3 sub-modules may
include flags, fragment, length, and binds several VCs into a VC Group be used: fragmentation, sequencing,
sequence number. At the destination (VCG) and transfers it on the same and timing. If a client signal exceeds
end—after terminating the LSP and T-LSP. In this way, the complexity of the maximum packet length that can be
withdrawing the outer T-LSP label—the the transport and switching devices of carried by a service layer network, it
device decides which high-layer the network is reduced and less must be fragmented. Some clients
service instance the data stream bandwidth is consumed. require sequenced delivery and
belongs to. This is based on the inner real-time support, so their signals
CII label. 3.1 Packet Switching and Forwarding should be delivered in a sequenced
The transport plane of MPLS-TP has The service packet switching and and timed way. A sequencing function
an additional feature: It is transparent to forwarding functions of MPLS-TP are provides three services: frame
the client layer and service layer. mainly involved in exchanging ordering, frame duplication detection,
Transparency to the client layer means MPLS-TP labels of the packet client and frame loss detection. SDH and FR
that any client signal can be carried on data and forwarding the data. client signals may require timed clock
MPLS-TP network and transmitted in Specifically, the functional modules recovery and timed packet delivery.
the form of packets. The client network include pre-processing, forwarding In an MPLS-TP network, the
can be IP, Ethernet, ATM, FR, PDH, or and switching, encapsulation, mapping from client signals onto link
SDH. Transparency to the service layer fragmentation, sequencing, timing, frames involves client service
means that MPLS-TP can use any multiplexing/demultiplexing, and encapsulation, signal multiplexing, and
underlying technology for transport.
Before MPLS-TP has its own data link VC Binding
layer protocol, packets can be carried
and transmitted via existing Ethernet or VC Label
SDH networks. VC Label
T-LSP Label Bound T-LSP Label
VC Label VC Label
VC Label
3 Label Forwarding
VC: Virtual Circuit T-LSP: Transport Label Switched Path
Mechanism
MPLS-TP adopts the dual-label ▲Figure 2. An example of MPLS-TP dual-label structure.
Service Adaptation and Label Forwarding Mechanism for MPLS-TP Special Topic
(600 in this example).
Multiplexing
ETH
Tunnel ×220 Terminal Device 1 converts the
MPLS-TP T-LSP label into T-LSP tunnel
LSP
Label Sequencing
Management
Type-Length-Value (TLV). the local
GFP-F Entry ×220
CII label into LABEL TLV, and the CII
Multiplexing
×1 Tunnel ×220
…
T-MPLS Identifier (CII-ID) into Forwarding
IP
Equivalence Class (FEC) TLV. It then
Header
PPP/ LSP
HDLC Shim
×1 sends a LABEL MAPPING message to
Sequencing no-IP Terminal Device 2.
Payload
Management Labe
Header
MPLS-TP
RPR ×1 Entry LSP Terminal Device 1 receives a LABEL
MAPPING message from Terminal
ETH: Ethernet LSP: Label Switched Path PPP: Point-to-Point Protocol Device 2 and decodes the message to
GFP-F: Generic Frame Procedure Framed MPLS-TP: Multiprotocol Label RPR: Resilient Packet Ring
HDLC: High Level Link Control Switching Transport Profile obtain CII label and CII-ID.
At the same time, Terminal Device 2
▲Figure 3. MPLS-TP mapping, multiplexing and fragment detection. runs the same steps (1) to (6)
independently.
mapping of MPLS-TP packets into link Terminal Device 1 of MPLS-TP network After the two terminal devices have
frames. Figure 3 illustrates the to establish an MPLS-TP VC to finished label binding, exchanging their
relationship between information units Ethernet VLAN 2. labels, and confirming their port
in an MPLS-TP network. Terminal Devices 1 and 2 of parameters are correct, T-MPLS VC
Client signals can also include OAM MPLS-TP network negotiate and with CII-ID 50 is successfully set up. If
packets of T-MPLS networks. Both the assign a VC Identifier (VCID) to the VC. one Ethernet LAN connection fails to
data packet and the OAM packets may Terminal Device 1 initializes a LDP work or is removed, a label cancel
have a label header for multiplexing. signaling session to Terminal Device 2 message will be sent to the opposite
Finally, MPLS-TP packets are mapped if one is not already present. Once the terminal device to cancel the assigned
onto data link frames, which are two terminal devices receive a LDP MPLS-TP CII label.
transmitted via MPLS-TP topological KEEPALIVE message from each other,
links. The forwarding and switching they set up the session and are ready 3.3 Dual-Label Transport Mode-based
module on MPLS-TP terminal devices to switch the labels bound on VPN Services
is responsible for switching the the VC. MPLS-TP networks have rich OAM
processed client data to related T-LSP After the Ethernet VLAN’ s state overhead functions[10] to monitor and
and forwarding them. Intermediate becomes“UP” , Terminal Device 1 manage signals in the network. This
transport switching devices continue to assigns a local CII label (500 in this improves the operability and security of
forward the packet data with MPLS-TP example) for the VC based on its VCID. the entire network, and paves the way
labels until the data is demultiplexed by It sets up T-LSP 1 for transport over the for Virtual Private Network (VPN)
the terminal device at the destination. VC, assigning the VC a T-LSP label services in the MPLS-TP network.
The destination terminal devices
forward the data to destination CEs[8-9]. Port A ， Port B ，
VLAN 100 Transmission VLAN 200
Opposite Terminal Switching Opposite Terminal
MPLS-TP Network
2.2.2.2. ， 1.1.1.1. ，
3.2 Signaling Transfer with Dual-Label CII-ID 50
Device
CII-ID 50
Transport Mode T-LSP 12
To explain signaling transfer in
dual-label transport mode, an example 600 Virtual Circuit 601
1A ，100 1B ，200
will be provided. Suppose there are 2 Terminal
Terminal
Ethernet Virtual Local Area Networks Device 1 Assign labels via indirect
Device 2
(VLANs): Ethernet VLAN 1 (port 1A, signaling session
VLAN 100) and Ethernet VLAN 2 (port 1.1.1.1 Transmission 2.2.2.2

Switching
1B, VLAN 200). An MPLS-TP VC is Device
established between them based on Assign label 501 to CII-ID 50
Label Distribution Protocol (LDP). Assign label 500 to CII-ID 50
Figure 4 illustrates a solution for Terminal device 1 chooses the tunnel to Terminal device 2 chooses the tunnel to 1.1.1.1.
2.2.2.2. The label is 600. The label is 601.
signaling transfer between the 2
Ethernet VLANs. 600 501 Ethernet 601 500 Ethernet
The process of establishing
Dual-Label Structure
MPLS-TP VC and transferring signals CII-ID: Common Interworking Indicators Identifier
MPLS-TP: Multiprotocol Label Switching Transport Profile
T-LSP: Transport-Label Switched Path
VLAN: Virtual Local Area Network
is described below.
Ethernet VLAN 1 makes a request to ▲Figure 4. Establishment of MPLS-TP VC based on LDP.
[6] ITU-T Rec G.8110.1/Y.1370.1. Architecture of

Transport MPLS (T-MPLS) layer network [S]. 2006.
MPLS-TP Domain [7] ZHANG Y J, ZHANG Z H, GU W Y, et al. A
VPN1 Customer Terminal Terminal Customer VPN3 collision-aware queue assignment scheme for
Edge 1 Device Device Edge 3 multiservice [C]//Proceedings of the IEEE
International Conference on Advanced Infocomm
Technology (ICAIT’09), Jul 7-9, 2009, Xi’an,
China. Piscataway, NJ, USA: IEEE, 2009.
[8] ITU-T New Supplement Y.Sup4. Transport
requirements for T-MPLS OAM and considerations
VPN2 Customer Terminal Terminal Customer
Edge 2
VPN4 for the application of IETF MPLS technology [S].
Device Device Edge 4
2008.
[9] ITU-T G.8121. Characteristics of transport MPLS
equipment functional blocks [S]. 2007.
[10] WARD D, BETTS M. MPLS architectural
T-LSP Label T-LSP Label considerations for a transport profile [R]. ITU-T
VPN Data Packet VPN Data Packet IETF Joint Working Team. 2008.
CII Label CII Label
VPN Data Packet VPN Data Packet
CII: Common Interworking Indicators T-LSP: Transport-Label Switched Path

MPLS-TP: Multiprotocol Label Switching Transport Profile VPN: Virtual Private Network
Biographies
▲Figure 5. Implementation of MPLS-TP VPN. Zhang Yongjun
Traditional VPN technologies (L2 or L3 adopted, where a T-LSP label identifies Zhang Yongjun, PhD, is an
tunneling technologies) are easy to the shared tunnel between terminal associate professor at Beijing
University of Posts and
implement, but are defective in devices, and a CII label identifies the Telecommunications (BUPT).
scalability, security, management and dedicated connection between CEs. His research interests include
maintenance, QoS, and traffic The CII label, in the form of a MPLS-TP broadband network access
and transfer technologies.
engineering. In particular, when a client label stack, is multiplexed in the Zhang has presided over
adopts multiple access technologies LSP-based tunnel of the MPLS-TP several projects of the National
High Technology Research
such as Point-to-Point Protocol (PPP), backbone network. The LSP can be and Development Program of
ATM, FR, and Ethernet, the operator regarded as a tunnel carrying several China “863”
( Program) and 5 projects of the
must configure several core networks VCs, and a VC is equivalent to a National Natural Science Foundation of China. Of
these, three have been awarded ministry-level
accordingly in order to offer VPN point-to-point route established by the Science and Technology Advancement Prizes. He
services. VPN services based on LDP for VPN users. has published over 40 papers and one book in
English.
MPLS-TP technology can be
implemented in a unified network
platform that is independent of 4 Conclusion Zhang Zhihui
transport technologies. With good service adaptation and label Zhang Zhihui is a doctoral
MPLS-TP VPN is more easily forwarding capabilities, MPLS-TP candidate at BUPT. His
achieved with point-to-point tunnels technology is well-suited for packet research interests include next
generation optical network
set up in the MPLS-TP backbone transfer. However, it is rivaled by other technologies—specifically, the
network. Its CEs do not participate in L3 packet transport technologies such as development and design in
traffic control and resource
routing, and the client itself configures PBT. As an emerging technology, allocation of optical
routing within the VPN. This makes MPLS-TP needs the joint efforts of networks—and survivability
MPLS-TP VPN independent of the standardization organizations, telecom technologies for next
generation packet transport
client’s L3 network protocol. In operators, and equipment vendors to networks.
MPLS-TP VPN, terminal devices are hasten its maturity.
only responsible for L2 connection, and Gu Wanyi
References
forwarding between CEs. Functions of [1] 朱京, 刘昭伟, 雷学义, 等. PTN-信息通信基础承载网
L3 or above are performed by CEs. The 络的演进与变革[J]. 电力系统通信, 2009, 30(4): Gu Wanyi is a professor and
MPLS-TP network itself can carry 52-59. doctoral adviser at BUPT. She
ZHU Jing, LIU Zhaowei, LEI Xueyi, et al. has been engaged in
different client signals. By means of PTN-evolution of information and communication researching and teaching
signal adaptation and encapsulation, a transport network [J]. Telecommunications for optical communications for
Electric Power System, 2009, 30(4): 52-59. many years. Her research
client can use different technologies to [2] NIVEN-JENKINS B, BRUNGARD D, BETTS M, et al. interests include high speed
access the MPLS-TP VPN, thus MPLS-TP requirements [R]. optical fiber communication
satisfying the client’s demand for low draft-jenkins-mpls-tp-requirements-01. 2008. systems, intelligent optical
[3] ITU-T G.8114. Operation and maintenance networks, and next generation
cost interconnection of VPNs of mechanisms for T-MPLS layer networks [S]. 2007. optical node devices. She has
different layers. [4] ITU-T Rec G.8131.1. Transport MPLS (T-MPLS) been awarded five ministry-level Science and
layer network protection switching [S]. 2007. Technology Advancement Prizes, and has
Figure 5 illustrates MPLS-TP VPN. It [5] ITU-T Rec G.8112. Interfaces for the Transport published over 100 academic papers.
can been seen that dual labels are MPLS(T-MPLS) hierarchy [S]. 2006.
PTN Clock Synchronization Technology and Its Applications Special Topic
Li Qin
PTN Clock Synchronization

Technology and Its Applications
Li Qin
(ZTE Corporation, Shenzhen 518055, P. R. China)
Abstract:
Clock synchronization is an important issue for packet transport networking. Current clock synchronization technologies include
synchronous Ethernet, IEEE 1588v2, and Network Time Protocol (NTP). However, individually, these technologies are beset with certain
problems. Synchronization Status Message (SSM) algorithm for synchronous Ethernet standards suffers clock ring, and has difficulty
tracing and counting nodes. An extended SSM algorithm can improve clock synchronization. NTP is too imprecise to meet the
requirements of telecom networks, yet IEEE 1588v2 alone can lead to slow convergence time and influence time delay precision when the
network is heavily loaded. ZTE therefore proposes an IEEE 1588v2 solution based on synchronous Ethernet in order to effectively raise
the precision of Packet Transport Network (PTN) time synchronization.
A
s operators seek to replace their PTN, including G.8261 (defining the physical links or obtain the clock
conventional Time Division general requirements), G.8262 from external synchronous interfaces. A
Multiplexing (TDM) transport (defining equipment clock node selects the best quality clock from
networks with Packet Transport performance), and G.8264 (defining among the multiple clock sources, and
Networks (PTN), clock synchronization system architectures and sets local clocks accordingly. Then the
comes into consideration. There are synchronization function modules). locked-in clock is transported to
two requirements for PTN IEEE 1588 standards were released equipment at lower layers. Each layer of
synchronization: in 2002 with the view of defining Precise the network becomes synchronized to
(1) It must be able to bear TDM Time Protocol (PTP) for a LAN multicast the Primary Reference Clock (PRC)
services and provide a clock recovery environment. through level-by-level locking. Similar
mechanism. However, standards released physical-layer synchronization
(2) PTN should have a specifically for this purpose could not technology can be used in packet
high-precision network reference clock be applied to telecommunications networks, as shown in Figure 1.
for synchronization of network nodes. environments with greater complexity.
IEEE 1588v2 was therefore issued in 2.1 Principle of Synchronous Ethernet
1 Overview of 2008 for application in telecom Synchronous Ethernet technology in
networks[1-5]. The IETF Network Time PTN uses Ethernet link steams to
Synchronization Protocol (NTP) implements time recover clocks. Ethernet physical layer
Technologies synchronization between users and coding adopts 4B/5B (FE) and 8B/10B
Clock synchronization includes between time servers in the Internet. (GE) technology, in which one
frequency synchronization and time additional bit is inserted into every four
synchronization. The former implies bits. In this way, four consecutive 1s or
identical timing intervals, while the latter
2 Synchronous Ethernet four consecutive 0s will not appear in
implies identical starting time as well as Technology any transported data stream, and the
intervals. Different wireless standards Physical layer synchronization clock information is effectively
have different requirements for clock technology is widely applied in contained. High-precision clocks are
bearing. traditional Synchronous Digital used to transport data from the Ethernet
Since 2004, the ITU-T Q13/SG15 has Hierarchy (SDH) networks. Each node source-end interface, and are
made a number of proposals relating to in an SDH can extract the clock from recovered and extracted at the
Special Topic PTN Clock Synchronization Technology and Its Applications
Li Qin
Packet Network oscillation;

•The algorithm is ideal for
Upper Layers Upper Layers Upper Layers processing because it supports
distributed processing at lower layers
PRC Physical Layer
Physical Layer Physical Layer and each NE is in an equal position;
•Due to the direct use of standard
S1 bytes, there is no problem
interworking with other vendors’
equipment.
: Timing Flow : Data Flow RRC: Primary Reference Clock
▲Figure 1. Physical-layer synchronization in PTN.

3 Time Synchronization
receiving end. for network clock synchronization. Technology
The principle of synchronous However, it has two weaknesses: Time synchronization technology is a
Ethernet is shown in Figure 2. A (1) It cannot satisfactorily handle the progression from frequency
high-precision clock is embedded into problem of synchronous clock ring. synchronization. By using the packet
the physical layer chip of equipment at (2) It has clock signal attenuation. protocol data unit to carry clock/time
the source side (Node A). This clock is With an increase of synchronous links, messages, PTN time synchronization
used by the chip to send data out. The drift caused by noise and temperature technology can synchronize the master
physical layer chip of equipment at the changes in synchronization allocation clock and slave clock. Figure 3 shows
receiving side (Node B) extracts this deteriorate timing reference signals. the basic principle.
clock from data streams. Precision is Consequently, the number of
not damaged in this process and synchronized Network Elements (NEs) 3.1 Network Time Protocol (NTP)
synchronicity with the source end is is limited on the same synchronous link, Before the advent of IEEE 1588v2, there
maintained. The principle of clock and standard SSM has difficulty tracing were three main types of PTN time
transport in synchronous Ethernet is and counting nodes. synchronization protocol: time protocol,
similar to that of an SDH network—the For PTN, ZTE proposes an improved daytime protocol and NTP. NTP is
clock is recovered from the Ethernet and extended SSM algorithm. ESMC fulfilled purely by low-precision
physical link. As a result, the quality of packet uses two Type-Lengh-Values software; the widely used NTPv3, for
recovered clock is not influenced by (TLVs) to transport SSM messages. example, achieves synchronization
link service traffic, and clock tree Following the ITU-T standards, the first precision of 10 ms. IETF is now
deployment and quality is the same as TLV transports original SSM byte proceeding with NTPv4 standards that
that of SDH/SONET. This satisfies the information at the synchronization support IPv6 and dynamic discovery
timing interface index of G.823. quality level. The second TLV is used servers. Synchronization precision of
for path protection. The improved NTPv4 is expected to reach a 10 μs
2.2 SSM Algorithm for Synchronous algorithm has the following advantages: level. However, the stability and
Ethernet •It fundamentally prevents clock precision of NTP still fails to meet the
Synchronization Status Message (SSM) ring; high requirements of telecom networks.
algorithm originates from SDH clock •It automatically selects the optimal
synchronization control. Its application (shortest) route in the event of multiple 3.2 IEEE 1588v2 Protocol
rules and clock selection algorithm clock paths;
conform to the ITU-T G.781 standard, •If there is a route to the master 3.2.1 Implementation Principle of IEEE
and control of synchronous Ethernet clock, NEs will trace the master clock 1588v2
retains SDH network characteristics. instead of entering a state of free IEEE 1588v2 unifies time and frequency
Moreover, based on the conventional
clock network, SSM supports Transmit Receive
High-Precision Phsyical-Layer 1 G/10 G Phsyical-Layer
synchronous Ethernet by building in Clock Chip Chip MAC
Ethernet Synchronous Message
System Clock
Line Clock
Channel (ESMC). As described in

G.8264, ESMC is a unidirectional
Master Clock External Clock PLL
broadcast protocol channel of the
Media Access Control (MAC) layer Slave Clock
used for transporting SSMs between TDM E1 Ethernet TDM
E1
different equipment. The equipment Equipment Equipment
A B
selects the best clock source based on
MAC: Media Access Control PLL: Phase-Locked Loop TDM: Time Division Multiplexing
SSM in ESMC messages.
Standard SSM algorithm is effective ▲Figure 2. Synchronous Ethernet.
Li Qin
PRC Recovered Reference Timing Signal IEEE 1588v2 timestamp treatment is

performed by hardware. The timestamp
PSN processor is located between the
Timestamp Timestamp physical layer and MAC layer, as shown
Master Processing in Figure 5.
Hardware timestamp treatment
compensates for the time taken by IEEE
: Time Stamp PRC: Primary Reference Clock PSN: Packet Switching Network
1588v2 protocol frame to pass through
▲Figure 3. Packet protocol synchronization mode. a protocol stack. Therefore, precision of
message delivery and timestamp
synchronization. In the future, it will be IPv4/IPv6, and User Datagram Protocol receiving at the port is guaranteed.
suitable for inter-office time-frequency (UDP), IEEE 1588v2 defines three basic (2) Node Buffering and Path Delay
transport of different platforms. It is clock types: Ordinary Clock (OC), IEEE 1588v2 defines E2E TC and P2P
capable of unidirectional frequency Boundary Clock (BC), and Transparent TC for compensating node buffering
transport using Time over Packet (ToP) Clock (TC). delay. There are two methods of
based on IEEE 1588v2 timestamp. It OC is a single-port device acting as transport path compensation: delay
also realizes time synchronization by either the master or slave clock. Only request-response, and peer delay.
using IEEE 1588v2 protocols. one master clock is permitted in a Delay request-response works with
Therefore, it is widely applied in PTNs. synchronous domain. Frequency, E2E TC. This TC marks timestamp
IEEE 1588v2 time synchronization is accuracy, and stability of the master treatment in the packet at ingress and
achieved by encoding time signals clock is directly related to the egress, while the slave completes all
through use of master and slave clocks. performance of the entire synchronous the calculations on time delay
Both master and slave time are network. PRC or GPS is generally compensation.
synchronized by bidirectional message adopted for network synchronization. Peer delay works with P2P TC. This
interaction as well as by employing Slave clock performance determines TC calculates time delay among
network symmetry and delay the timestamp precision and sync endpoints, and each endpoint interacts
measurement technology. message rate. with the TC respectively to compute
According to the 1588v2 principle As a multi-port device, BC can P2P time delay. Based on these
shown in Figure 4, Delay= (T2-T1+ connect with many OCs/TCs. Among calculations, the slave then calculates
T4-T3)/2; Offset= (T2-T1-T4+T3)/2. the multiple ports of BC, one serves as delay compensation.
Messages such as Sync, Follow_Up, the slave port, connecting to the master
Delay_Req, and Delay_Resp are port of the master clock or other BCs. 3.2.4 Fulfillment of IEEE 1588v2 in PTN
delivered between the master and The other ports serve as master ports, IEEE 1588v2 synchronization precision
slave clocks. Referring to the four connecting to the slave port of the slave is affected by many factors in real
values T1, T2, T3 and T4, the delay and clock or BC at the next layer. They can network deployment. Its application in
offset between the master and slave also serve as backup ports. complex network environments, such
can be determined. TC is used for connecting the master as hybrids of microwave and switching
Synchronization messages include clock with the slave clock. It forwards networks, is still in the research stage.
general messages and event interacted synchronous messages
messages. A general message, such between the master and slave clocks in
as Follow_Up, does not handle a transparent way. It also calculates Master Slave
timestamp itself; it can carry accurate local buffering treatment time for
delivering/receiving time of event synchronous messages—such as Sync T1
messages such as Sync. In addition, it and Delay_Req—and writes this time Sync [a ]
implements network configuration and into the CorrectionField byte block of T2=T1+Delay+Offset
management, and fulfills the sync messages. The slave clock Follow_up [T1 ]
communication among PTP nodes. An calculates delay and offset based on

event message needs to deal with byte values and sync message
timestamp, and can carry timestamp or timestamp in order to realize T3
Delay_Req
not. Depending on the event message synchronization. TC can be divided into
timestamp or general message Edge-to-Edge (E2E) TC, and
timestamp, the slave clock can Peer-to-Peer (P2P)TC.
T4
determine the path delay and offset Delay_Resp[T4]
between the master and slave clocks. 3.2.3 Delay in IEEE 1588v2
Delay is a major factor affecting T4=T3+Delay-Offset
3.2.2 Clock Types precision of an IEEE 1588v2 system.
Based on protocols such as Ethernet, (1) Timestamp Treatment Delay ▲Figure 4. Principle of IEEE 1588v2 protocol.
Special Topic PTN Clock Synchronization Technology and Its Applications
Li Qin
entire network. Tree networking does

IEEE1588 Code not support clock routing protection.
Protocol Stack & OS However, in ring networking, when
current clock routing fails, related NEs
MAC
can receive messages such as warning
Timestamp Processor and SSM, and trace the source clock
PHY from other directions in order to protect
◀Figure 5. clock routing. Figure 6 shows examples
MAC: Media Access Control
Network IEEE 1588v2 timestamp of synchronous Ethernet networking.
OS: Operating System
treatment. Jittering of synchronous messages is
intensified after they have been
IEEE 1588v2 can achieve precision of synchronous Ethernet physical layer. transported by NEs. Therefore, optimal
100 ns in a pure PTN testing network. Moreover, it reduces the frequency of clock quality is achieved when network
However, due to the complexity of IEEE 1588v2 message packet delivery. equipment is able to trace the clock
network time delay and lack of control Time precision is not affected even if source using the shortest possible
over IEEE 1588v2 bidirectional path the network is heavily loaded, which path. ZTE’ s PTN adopts an extended
asymmetry, there are unpredictable improves reliability and precision of SSM algorithm which includes the
risks when relying solely on IEEE PTN time synchronization. To solve the number of nodes the clock has passed
1588v2 protocol and analytic algorithm engineering problem in PTN asymmetry through. In any case, NEs are able to
to adapt to real networks. When a measurement, ZTE configures its trace the clock source using the
network is heavily loaded, frequently access-layer PTN equipment with a shortest possible path.
sent IEEE 1588v2 message packets are time error measurement function that NE C can use point B or D to trace
easily influenced by service packets, quickly, easily, and accurately performs clock information with its source at A.
and this greatly impacts time delay measurements without the need for The clock passes through only one
precision. However, by decreasing the professional instruments. node when tracing from point B, and
frequency of sending message two nodes when tracing from point D.
packets, the time convergence rate To obtain higher clock quality, ZTE’ s
may be slowed. In addition, the IEEE 4 Typical Applications PTN automatically chooses the clock
1588v2 algorithm must be traced from B.
compensated with bidirectional path 4.1 Application of Synchronous Ethernet
asymmetry—achieved by optical Similar to SDH, synchronous Ethernet 4.2 Applications of IEEE 1588v2
fiber—in actual network construction. supports both ring networking and tree
Timing error of optical fiber asymmetry networking. Generally, RNC provides 4.2.1 Replacing GPS at Base Stations
is first measured by costly time the clock source, and clock information Typical IEEE 1588v2 application
synchronization testers and arrives at every base station via involves replacing GPS at Radio
oscilloscopes, and is followed by synchronous Ethernet in order to Access Network (RAN) base stations.
asymmetrical time delay compensation. maintain synchronicity throughout the Installing a GPS antenna at a
Owing to the large number of PTN
nodes, compensation work requires
many engineers. Instruments such as
time synchronization testers and RNC
oscilloscopes are also cumbersome. It RNC
is therefore difficult to apply IEEE

1588v2 widely in real network FE
construction, and opinions still vary
about its feasibility. Node B
ZTE Corporation proposes a Node B
synchronous Ethernet-based IEEE FE
FE E1
1588v2 time transport solution to the Node B
problem. The core idea is to establish a Node B
Node B
highly controllable network where
clocks are isolated from time, Node B FE
E1
eliminating unpredictable risk. This is E1
Node B
beneficial for fast convergence of time Node B Node B
Node B
synchronization in order to implement RNC: Radio Network Controller
IEEE 1588v2 based on steady
frequency synchronization of the ▲Figure 6. Examples of synchronous Ethernet networking.
Li Qin
act as the GPS backup) and offer In this case, IEEE 1588v2 can be used
higher security guarantee for RAN. to obtain time frequency in these
Figure 8 shows an example of IEEE ordinary networks.
A 1588v2 replacing GPS at base stations. Figure 9 shows an example of
Only one NE in the PTN needs to frequency recovery 1588v2 networking.
receive time messages from GPS; for When the network between equipment
E example, through the 1PPS+ToD A and equipment B is an ordinary data
B interface. PTN delivers the time network, IEEE 1588v2 Sync messages
messages to other NEs by IEEE 1588v2 are transported from A to B via the
protocol, then the messages arrive at ordinary data network. Equipment B
D the base stations via Ethernet or other uses IEEE 1588v2 to recover the clock.
interfaces. Time synchronization among The recovered clock is taken as the
all the base stations is fulfilled reference source of B, with which the
C accordingly. service clock is recovered.
▲Figure 7. An example of clock tracing. Base stations support either IEEE
1588v2 protocol or time interfaces. If
TD-SCDMA or CDMA2000 base station supporting IEEE 1588v2 protocol, PTN 5 Conclusions
requires an area of 120 degrees to be works in TC mode; if not, PTN works in Research into PTN clock
cleared, and this can be difficult given BC mode. synchronization technology is
environmental constraints. GPS is deepening. ZTE proposes an extended
difficult to install indoors and 4.2.2 Frequency Recovery synchronous Ethernet SSM algorithm
underground. Moreover, GPS is IEEE 1588v2 is also used to recover and IEEE 1588v2 solution based on
expensive and has a high failure rate. If frequency by ToP. Most operator synchronous Ethernet. This solution
PTN can provide time synchronization networks are ordinary data networks, plays an important role in raising the
for base stations, it can replace GPS (or not supporting synchronous Ethernet. precision of time synchronization in
PTN and reducing the level of
GPS GPS engineering complexity. PTN clock
synchronization technology has
extensive application possibilities,
especially in RAN, TDM services, M2M
real-time data acquisition, and VIP
private networks.
PTN PTN References

[1] ITU-T G.8261.Timing and synchronization aspects
in packet networks [S]. 2006.
PTN [2] ITU-T G.8262.Timing characteristics of
synchronous Ethernet equipment slave clock (EEC)
PTN [S]. 2007.
[3] ITU-T G.8264. Distribution of timing through packet
networks [S]. 2008.
[4] IEEE 1588.IEEE standard for a precision clock
synchronization protocol for networked
Node B Node B measurement and control systems [S]. 2008.
[5] MEF 22. Mobile backhaul implementation
GPS ：Global Positioning System PTN ：Packet Transport Network agreement phase1 [S]. 2009.
▲Figure 8. An example of IEEE 1588v2 replacing GPS at base stations.
PRC PRC Biography

Li Qin
Differential Timeing Messages
Li Qin received his master’s

degree from Beijing University of
Posts and Telecommunications.
B He is a PTN solution manager at
A PTN the Bearer Network Planning
◀Figure 9.
System Department of ZTE, and
is mainly engaged in the study
Networking of of 3G/LTE PTN bearer solutions.
frequency recovery by
PRC: Primary Reference Clock PTN: Packet Transport Network
IEEE 1588v2.
Special Topic PTN and IP-Based Mobile Backhaul
Cong Kai, Zhao Fuchuan
PTN and IP-Based Mobile Backhaul

(Bearer Network Planning & System Design Department, ZTE Corporation, Shenzhen 518055, P. R. China)
Abstract:
Packet Transport Networks (PTN) combine the advantages of packet technology and Synchronous Digital Hierarchy (SDH) technology.
Taking the packet switch as its core, PTN has a high statistical multiplexing ability, which allows it to become a more efficient packet
transfer service. Its strong Operation Administration and Maintenance (OAM)—similar to SDH and carrier-class security protection
services—ensures efficient mobile backhaul business management and transmission quality. MPLS Transport Profile (MPLS-TP)
packet transport network eliminates connectionless features, such as Penultimate Hop Popping (PHP), label merge and Equal-cost
multi-path (ECMP), and is enhanced in terms of OAM, protection, and synchronization. This is ideally suited for carrying IP-based
mobile backhaul services and key account services. PTN, original Multi-Service Transport Platform (MSTP), Metro Ethernet, and IP over
Wave Division Multiplexing (WDM)/Optical Transport Network (OTN) of organic complexes, contribute to an "all IP" era of
telecommunications services.
W
ith the rapid development of Figure 1 illustrates the evolution of much more bandwidth is required for
mobile communication mobile communication technologies. It transport networks, and transmission
technologies, 3G and Long can be seen that uplink and downlink efficiency must be improved in order to
Term Evolution (LTE) systems bandwidths and rates have reduce costs.
exist not just on paper, but are substantially increased during the In the 2G era, leading technologies of
beginning to be deployed in reality. evolution from 2G to 3G/LTE. This mobile backhaul networks included
While the development of mobile increase in interface rate means that Synchronous Digital Hierarchy (SDH)
technologies has bright prospects
for the future, it also means new GSM GPRS EDGE E-EDGE
challenges for mobile backhaul LTE
networks.
The trend towards all-IP TD-
HSPA
MC-HSPA
LTE
telecommunications services is SCDMA MBMS (TDD)
reflected across the whole mobile
communications industry. Services
WCDMA HSDPA HSUPA HSPA+
are being transitioned away from R99/R4 R5 R6 R7/R8
LTE
Time Division Multiplexing (TDM)
DL:384 kbit/s DL:14.4 Mbit/s DL:14.4 Mbit/s DL:28.8/84 Mbit/s DL:100 Mbit/s
voice to IP-based voice and data. UL:384 kbit/s UL:384 kbit/s UL:5.76 Mbit/s UL:11.5 Mbit/s UL:50 Mbit/s
Because IP technologies are being CDMA 2000 EV-DO EV-DO
widely used in 3G networks and CDMA 1x CDMA 2000
EV-DO Rev.A Rev.B
LTE
mobile data services continue to

DL:14.4/64 kbit/s DL:153.6 kbit/s DL:2.4 Mbit/s DL:3.1 Mbit/s DL:6.2-73.5 Mbit/s DL:140-280 Mbit/s
grow, pressure is mounting on 3G UL:14.4/64 kbit/s UL:153.6 kbit/s UL:153.6 kbit/s UL:1.8 Mbit/s UL:3.6-27 Mbit/s UL:34-68 Mbit/s
mobile backhaul networks for better
service sensing, higher Quality of 2G 2.5G 2.75G 3G 3.5G 3.75G 3.9G 4G
Service (QoS), and more efficient
DL: Downlink HSPA: High Speed Packet Access MC-HSPA: Multi-Carrier HSPA
statistical multiplexing. The change EDGE: Enhanced Data Rates for GSM Evolution HSUPA: High Speed Uplink Packet TD-SCDMA: Time Division Synchronous Code
of interface from E1 to FE has also E-EDGE: Evolved-EDGE Access Division Multiple Access
led to a sharp increase in interface EV-DO: Evolution-Data Optimized LTE: Long Term Evolution UL: Uplink
GPRS: General Packet Radio Service MBMS: Multimedia Broadcast and WCDMA: Wideband Code Division Multiple Access
bandwidth. In LTE systems, base HSDPA: High Speed Downlink Packet Access Multicast Service
stations may even adopt a GE
interface with a rate of 1,000 Mbit/s. ▲Figure 1. Evolution of mobile communication technologies.
PTN and IP-Based Mobile Backhaul Special Topic
Area Network (LAN) technologies. It is

Emulated Service
based on Media Access Control
PW
(MAC)-in-MAC[3] specified in IEEE
PSN Tunnel
802.1ah. In PBB-TE, the MAC address
self-learning function is disabled, but
PWES
PE1 PE2
PWES network management and control
PW1 functions are enabled. At present,
CE1 CE2 MPLS-TP is the de facto mainstream
PW2 technology for PTNs.
MPLS-TP standardization is led by
the International Telecommunication
Native Service Native Service
Union-Telecommunication
CE: Customer Edge PSN: Packet Switched Network PWES: Pseudo Wire Emulation Service Standardization Sector (ITU-T) and the
PE: Provider Edge PW: Pseudo Wire
Internet Engineering Task Force (IETF).
▲Figure 2. Edge-to-edge services in PTN. These two organizations established a
Joint Working Team (JWT) in February
and Multiservice Transport Platform capabilities. 2008, and have made great progress in
(MSTP), both designed to aggregate PTN takes connection-oriented MPLS-TP standardization. As at
and efficiently transmit TDM-based packet technologies as its core, and February 2009, the JWT had worked out
circuit services. In particular, MSTP was adopts the abovementioned five RFC documents, two
designed to enable IP services to be transmission characteristics. As an recommendations, and 13 drafts. Key
carried over transport networks. It is integrated transport technology, it is standards of MPLS-TP are expected to
not, however, an all-IP solution. Its user used mainly to carry carrier-class be released in 2011.
interface may be IP-based but at its Ethernet services as well as TDM and
core, it is still TDM circuit-switching Asynchronous Transfer Mode (ATM)
and adopts a fixed tunnel for carrying services. Its packet core uses 2 Key Technologies of PTN
packet services. As a result, MSTP bandwidth efficiently and provides PTN is an integrated transport network
suffers defects such as low flexible tunnels with powerful statistical technology featuring both packet and
transmission efficiency, high cost, and multiplexing capability. So PTN is transmission. It has become a
poor scalability when carrying and particularly suitable for packet services mainstream solution for IP-based
transmitting variable-length packets, which are characterized by bursts. It mobile backhaul networks in 3G/LTE
bursty IP, and Ethernet services. In inherits SDH-like transmission systems due to several key
short, SDH and MSTP cannot meet the characteristics, powerful OAM, technologies.
requirements of 3G/LTE systems, and carrier-class protection, and Graphical
they restrain the development of mobile User Interface (GUI) network 2.1 Pseudo Wire Emulation
services. Packet Transport Network management. As a result, users feel a Edge-to-Edge (PWE3)
(PTN)—integrating packet technologies sense of continuity and consistency 3G systems are developing at pace but
and SDH—is therefore introduced. with traditional mobile backhaul traditional TDM services will continue to
networks[1]. be the main source of profit for telecom
Two technologies are currently operators for some time. Therefore, PTN
1 Technical Features of available for PTNs: Multiprotocol Label should be capable of carrying multiple
PTN Switching Transport Profile services. PWE3 is designed specifically
3G/LTE systems can be broadband (MPLS-TP)[2] and Provider Backbone for this purpose. By using PWE3 circuit
and IP-based broadband. To Bridge Traffic Engineering (PBB-TE). emulation, MPLS-TP adapts itself to all
accommodate these, a mobile MPLS-TP is an extension of core types of customer services—including
backhaul network must have traditional network technologies. Based on Ethernet, TDM, and ATM—and
capabilities such as efficient statistical Multiprotocol Label Switching Transport provides dedicated edge-to-edge
multiplexing, flexible service sensing, (MPLS) of IP-based core networks, it tunnels for these services.
and differentiated QoS. As a simplifies the transport plane and PWE3 is a service emulation
carrier-class service bearer network, it complex control protocols. Moreover, mechanism which encapsulates
should also inherit the transmission by eliminating connectionless features, service data in a special circuit
characteristics of old systems, it enhances OAM and protection emulation header. This header includes
including end-to-end service switching mechanisms, guarantees the basic attributes of service data such
management, hierarchical Operation, reliable QoS, and provides statistical as frame format, alarm, signaling, and
Administration and Maintenance multiplexing. PBB-TE is a synchronous timing[4]. Edge-to-edge
(OAM), and carrier-class protection. connection-oriented transport services in the PTN are illustrated in
PTN technology has all these technology and an expansion of Local Figure 2. In PWE3, a Pseudo Wire (PW)
supports ring, linear, and mesh

Section Layer
Layer 2
Label 14 OAM Packet Content protection modes. Each mode has its
Encapsulation
own merits and shortcomings, suitable
for different scenarios. Ring protection
Layer 2 can provide a high level of protection
LSP Layer LSP Label Label 14 OAM Packet Content
Encapsulation specifically for ring topology[7], while
linear protection is not specific to any
topology and can be implemented with
Layer 2
PW Layer
Encapsulation
LSP Label PW Label Label 14 OAM Packet Content two connections for each path in a fixed
network. In application, these modes
LSP: Label Switched Path OAM: Operation, Administration and Maintenance PW: Pseudo Wire are often used in combination in order
to achieve carrier-class service
▲Figure 3. OAM packet encapsulation formats specified in ITU-T G.8114. protection.
OAM and protection are closely
is set up and maintained in a Packet this mechanism can quickly locate a related to each other. Using the OAM
Switched Network (PSN) tunnel. A PTN fault, and detect network mechanism to achieve timely fault
Provider Edge (PE) uses the PW to performance issues such as packet detection is a premise for carrier-class
encapsulate service data for loss rate and delay[5]. protection. There are three types of
transmission, keeping the attributes ITU-T specifies the OAM packet OAM related to the PTN protection
and features of the service as much as encapsulation formats[6] for Transport mechanism: alarm-related,
possible. For a Customer Edge (CE), Multiprotocol Label Switching performance-related, and
the PW is merely a link or circuit (T-MPLS), as shown in Figure 3. The communication channel-related.
occupied exclusively by a specific OAM signaling packets of each layer
service. This circuit is called virtual also contain labels encapsulated with 2.4 Synchronization
circuit. The CE does not recognize the MPLS. To distinguish OAM signaling The development of mobile
existence of the core network and treats packets from users’service data communication technologies also has
all services it processes as native packets, a special label (14) is defined. implications for time synchronization in
services. By defining a series of OAM protocol mobile backhaul networks. ITU-T,
packets, G.8114 implements diverse IEEE, and other standardization
2.2 QoS OAM functions. organizations have proposed numerous
In a SDH-based mobile backhaul PTN’ s all-round carrier-class time synchronization solutions. Among
network, an exclusive rigid tunnel is protection is illustrated in Figure 4. PTN them, the most attractive is precision
provided for all services. This provides access link protection, clock synchronization protocol
transmission mode is highly reliable, network-level protection, and suggested in the IEEE 1588v2
but on the other hand leads to device-level protection. It also standard[8]. This protocol adopts a
transmission waste because services
may have different characteristics.
Voice services, for example, with high
real-time characteristics, have different BTS/Node B 三三
三三
BSC/RNC
transmission requirements to common
Internet services. PTN can sense
e-NB
AGW 三三
三三
service characteristics and provide SR/BRAS
transmission as required. For telecom
operators, offering suitable
transmission for different service flow is
Access Link Protection: Network-level Protection: Device-Level Protection:
cost-efficient, especially when the ■ LACP Protection ■T-MPLS/MPLS-TP Path Protection ■ 1+1 Hot Backup for Clock,
demand for bandwidth increases ■ IMA Protection ■T-MPLS/MPLS-TP PW Protection Switching and Control Boards
■T-MPLS/MPLS-TP SNCP Protection ■ 1+1 Hot Backup for Power and,
substantially. ■T-MPLS/MPLS-TP Ring Protection Fan boards
■ TPS Protection
2.3 Hierarchical OAM and Carrier-Class
Protection
The OAM mechanism of PTN basically AGW: Access Gateway MPLS-TP: Multiprotocol Label Switching SR: Service Router
BRAS: Broadband Remote Access Server Transport Profile T-MPLS: Transport Multiprotocol Label
derives from SDH, and provides alarm BSC: Base Station Controller PW: Pseudo Wire Switching
and performance management for the BTS: Base Transceiver Station RNC: Radio Network Controller TPS: Tributary Protection Switching
IMA: Inverse Multiplexing over ATM SNCP: Subnetwork Connection
section layer, Label Switched Path LACP: Link Aggregation Control Protocol Protection
(LSP) layer, and Pseudo Wire (PW)
layer. By supporting hierarchical OAM, ▲Figure 4. All-around carrier-class protection of PTN.
PTN and IP-Based Mobile Backhaul Special Topic
master/slave clock configuration and

encodes the time before transmission.
Taking advantage of the symmetry of
RNC
三三
三三三三
三三三三
三三三三
三三三
三三三三三
三三
network links and delay measurement
technology, it synchronizes frequency,
Core Layer 三三
三三三三
三三三三
三三三三
三三三
三三三三三
三三
phase, and the absolute time of master
and slave clocks.
ZTE has been the first to adopt IEEE WDM/OTN
1588v2 Precision Time Protocol for
clock synchronization based on
synchronous Ethernet[9]. The core idea
of this solution is to set up a network Aggregation 10 GE 10 GE 10 GE
with separate but highly controllable Layer
clock time in order to avoid
unpredictable risks. With synchronous
Ethernet at the physical layer,
frequency synchronization is Access Layer GE GE 10 GE
implemented among nodes, thus
controlling the nodes’phases within a OTN: Optical Transport Network RNC: Radio Network Controller WDM: Wave Division Multiplexing
certain range and guaranteeing
accurate frequency. IEEE 1588 protocol ▲Figure 5. Flexible networking of PTN services with WDM/OTN.
is only used for phase tuning and time
transfer. A clock dedicated network without being affected during 3G access for customers who demand
constructed in this way is construction. Second, 3G services can small bandwidth but high security and
connection-oriented, strictly develop smoothly, bandwidth demand privacy, or by covering regions
controllable, and point-to-point. It of future data services can be easily temporarily out of reach of PTN.
avoids interference from asymmetrical satisfied, and mobile networks can
IEEE 1588 systems during deployment, evolve gracefully to LTE. This solution 3.2 PTN vs. WDM/OTN
and its time synchronization provides a clear structure of the mobile IP over Wave Division Multiplexing
performance is independent of network backhaul network, greatly facilitating (WDM)/Optical Transport Network
loader traffic. future planning, management, and (OTN) implements flexible dispatching
maintenance of the network. and protection of large-granule
3 Relationship between Depending on the availability of services based on wavelength or
MSTP resources in the existing network, Optical Cannel Data Unit (ODU). When
PTN and Other Bearer a networking mode comprising a constructing a Metropolitan Area
Networks mixture of PTN and MSTP can be Network (MAN), IP over WDM/OTN can
adopted in some areas of the network be applied in the core layer, providing
3.1 PTN vs. MSTP Netowrk as a supplement to the above solution. networking, dispatching and protection
In the 2G era, SDH/MSTP-based That is to say, PTN devices can be of large-granule services of core
networks provided stable and reliable introduced into the network access network elements. If a large MAN is
transmission for mobile services and layer or aggregation layer of existing used to carry mobile backhaul services,
were widely applied in mobile networks. MSTP network where there is PTN + WDM/OTN networking mode can
In 3G/LTE systems where PTN has bandwidth shortage. This not only be used. In this mode, the
been introduced, the issue arises of implements cross-area resource access/aggregation layer is networked
handling the relationship between PTN dispatching, but also enables the in the form of PTN—which dispatches
and the existing MSTP network. Taking operator to accumulate experience in services upwards to the PTN core
into account network complexity, low cost PTN planning, operation, dispatching layer in the Radio Network
construction expenses, and operation maintenance, and management. Controller (RNC) equipment room via IP
and maintenance costs, an In short, 2G networks will continue to over WDM/OTN. This greatly simplifies
independent PTN plane is able to be be highly profitable for operators for the network that contains core layer
set up in addition to the existing MSTP quite some time. This means that MSTP nodes and aggregation layer backbone
plane. PTN must co-exist. However, in the nodes, and flexibly dispatches services
Although this requires a large long term, after all mobile during adjustment of service homing.
investment in the early stages of PTN communication networks have become Moreover, as the core layer nodes are
construction, the advantages are IP-based, MSTP will inevitably be only connected to the RNC equipment
significant. First, existing 2G services replaced by PTN[10]. It may even room, the problem is avoided whereby
can be carried by the MSTP network supplement PTN in providing dedicated all node devices must be upgraded
because one node has been upgraded become vulnerable. mobile backhaul. The introduction and
to expand its service capacity (a development of PTN will, in turn, drive
problem sometimes encountered in the development of service networks
pure PTN networks). Network costs are
4 Application of ZTE’
s PTN and promote the rapid growth of
therefore saved. WDM/OTN provides Products IP-based services.
flexible dispatching and networking for With strength in research and
PTN services, as shown in Figure 5. development, and with significant References
[1] 武向军. 承载技术的革命—PTN [J]. 通信世界, 2009,
investment, ZTE now provides a full 40 (5): 11-15.
3.3 PTN vs. Metro Ethernet range of high-performance PTN WU Xiangjun. Revolution of bearing
In constructing a MAN, and depending products. In bidding tests conducted technology—PTN [J]. Communications World, 2009,
40 (5): 11-15.
on the nature of services carried on the by China Mobile in 2009, ZTE’ s [2] IETF RFC 5317. JWT report on MPLS-TP
network, an independent approach is products ranked first. Moreover, ZTE architectural considerations [S]. 2009.
[3] IEEE 802.1Qay. IEEE standards for local and
often adopted. Two planes are has deployed its PTN products in metropolitan area networks: virtual bridged local
constructed: a high-value mobile systems of several multinational area networks, amendment 7: provider backbone
backhaul network and a low-value operators including Telefonica, Telenor, bridge traffic engineering [S]. 2009.
[4] IETF RFC3985. Pseudo wire emulation
broadband access network. PTN is and TIM. Edge-to-Edge (PWE3) architecture [S]. 2005.
primarily used for carrying high value In the second quarter of 2009, China [5] HE Tingzong. OAM technology of packet transport
network [J]. ZTE Communications, 2009, 7(1):
services such as mobile backhaul and Mobile conducted performance tests 33-35.
services for important group customers. on PTN products and IP-based Iub [6] ITU-T Recommendation G.8114. Operation &
maintenance mechanism for T-MPLS layer
Low-value services such as public interfaces of several vendors. These networks [S]. 2007.
broadband access and services for tests were conducted in its telecom [7] ITU-T Recommendation G.8132. T-MPLS shared
large customers are first carried over a network in Shenzhen and were protection ring [S]. 2007.
[8] IEEE Std. 1588-2002. IEEE standard for a precision
Passive Optical Network (PON), and designed to evaluate PTN products in clock synchronization protocol for networked
are then sent to the service control terms of multi-service bearing measurement and control systems [S]. 2002.
[9] ITU-T Recommendation G.8261. Timing and
layer, Broadband Remote Access capability, QoS, OAM, time synchronization in packet networks [S]. 2006.
Server (BRAS) or Service Router (SR) synchronization, and longevity. When [10] 龙文富. 聚焦 TCO 构建 PTN 新型分组承载 [J]. 通信世
through a metro Ethernet. The two ZTE’ s PTN products were used to carry 界, 2009, 40 (5): 16-19.
LONG Wenfu. New type packet bearing in PTN:
planes work independently and do not services via IP-based Iub interfaces, lowering TCO [J]. Communications World, 2009,
interfere with each other. Such a service indexes and transmission 40 (5): 16-19.
strategy takes the following into quality met requirements. Compared
consideration: with MSTP products, ZTE’ s PTN even Biographies
(1) Important services such as excelled MSTP in some indexes and Cong Kai
Internet services and mobile backhaul satisfied the TD service requirements of
services differ in configurability and China Mobile. Cong Kai received his master’s
have different reliability requirements. A To date, ZTE’ s PTN products have degree from Beihang
shared plane that carries all these been used in China Mobile’ s PTN University, and currently works
as a planning manager for
services creates greater network network construction in over 20 carrier network products at ZTE
complexity. Moreover, Internet services provinces. Corporation. He has been
engaged in researching optical
require much more bandwidth than communication technologies for
mobile backhaul services. If they share many years.
the same plane, MAN access devices 5 Conclusions
must be re-configured for large Service requirements are the driving
capacity, high reliability technologies, force behind the development of PTN Zhao Fuchuan
and this increases costs. technologies. In the evolution to 3G,
(2) Broadband services are subject High Speed Packet Access Plus Zhao Fuchuan received his
doctoral degree from
to frequent change. So sharing the (HSPA+), and LTE networks, traditional Shanghai Institute of
same plane with mobile backhaul mobile services must become Microsystem and Information
services may affect the stability of IP-based because emerging mobile Technology (SIMIT) of the
Chinese Academy of
mobile services. If the stability of mobile services impose greater requirements Sciences. He is chief engineer
services is to be guaranteed, demand on synchronization, network delay, at ZTE Corporation,
specializing in PTN products.
for broadband services cannot be reliability, and security. PTN is He has been engaged in
quickly accomodated. Stability and introduced in this context. As the researching optical
quick response to service demands industrial chain matures, PTN will communication technologies for many years. In
2004, ZXMP—a product which Zhong oversaw
cannot be realized at the same time. advance technologically, have greater research and development—was awarded first prize
(3) Internet services are open. If they cost benefits, and will eventually at the Science and Technology Progress Awards of
Guangdong Province. At present, he is researching
share the same plane as mobile become the mainstream carrier and planning PTN products.
backhaul services, the network may platform for high value services such as
Analysis of Time Synchronization in PTN Special Topic
Li Han
Analysis of Time Synchronization

in PTN
Li Han
(China Mobile Research Institute, Beijing 100053, P. R. China)
Abstract:
For Time Division Synchronous Code Division Multiple Access (TD-SCDMA) and Time Division Long Term Evolution (TD-LTE) wireless
systems, using Global Positioning System (GPS) for time synchronization is problematic. In these wireless systems, GPS is costly,
insecure, and difficult to deploy. Nowadays, transportation of high precision time/phase synchronization signals via fiber, and based on
Precision Time Protocol (PTP) has become mainstream technology. This paper analyzes the main factors affecting time synchronization in
a Packet Transport Network (PTN) adopting IEEE 1588v2. Laboratory experiments and field tests prove the feasibility of transporting high
precision time synchronization signals through a PTN adopting IEEE 1588v2. A comparison is also made between different networking
modes for a PTN adopting IEEE 1588v2.
W
ith the development of mobile technology in the future. networks, further research must be
telecommunications Terrestrial transmission of high undertaken into precision, networking
technologies, precise network precision time synchronization signals mode, maintenance, and Best Master
time synchronization has involves two key technologies: Clock Algorithm (BMCA). This paper
become increasingly important. Precision Time Protocol (PTP), and analyzes the factors that may affect
CDMA2000, Time Division Synchronous compensation technology for time synchronization precision in a
Code Division Multiple Access transmission delay and jitter. Networks Packet Transport Network (PTN)
(TD-SCDMA), and Time Division Long using Network Time Protocol (NTP) adopting IEEE 1588v2[1-5].
Term Evolution (TD-LTE) base stations synchronization are only accurate to
all require high precision time dozens of milliseconds, which does not 1 Key Factors Affecting
synchronization. In TD-SCDMA meet the requirements of TD-SCDMA
systems, for example, the time systems. Based on a delay
Time Synchronization
synchronization index is ±1.5 μs. Such request-response mechanism, IEEE Precision
precision cannot be achieved through a 1588v2, also known as Precision Time IEEE 1588v2 adopts a master/slave
free running oscillator, or even through Protocol (PTP), uses synchronization clock configuration. The master clock
the frequency synchronization network. messages to calculate the time error distributes clocks periodically, and the
On the other hand, installing a Global between the slave and master clocks. receiver—taking advantage of network
Positioning System (GPS) for each Time Moreover, by using a link symmetry—measures time offset
Division (TD) base station is difficult to hardware-embedded time stamp as and delay. Synchronization of
engineer, costly, and insecure. well as a Boundary Clock (BC) or frequency, phase, and absolute time of
Therefore, time synchronization Transparent Clock (TC) to compensate master and slave clocks is thereby
protocol has become an important delay and jitter incurred by network realized. Transmission of IEEE 1588v2
technology for transmitting high components or protocol stacks, IEEE packets may require each node to
precision time synchronization signals 1588v2 achieves precision of process a time stamp, which incurs
in fiber systems. sub-microseconds. delay and jitter. As a result, the number
Using synchronization protocol to IEEE 1588 is a PTP originally of hops in a transport network impacts
transport high precision time designed for synchronization between the precision of time synchronization
synchronization signals via optical fiber industrial computers. Before it can be signals. Because a
systems is expected to be a leading applied to large-scale communication store-and-forwarding mechanism is
Special Topic Analysis of Time Synchronization in PTN
Li Han
GPS lengths of 64 bytes, 576 bytes, and

1􀆯518 bytes. The peak-to-peak timing
DUT1 DUT2 DUTn differences in each case are 250 ns,
GE GE 450 ns, 3􀆯200 ns, and 10 μs
respectively. It can be seen that the
1PPS+ToD
heavier the load or longer the packet,
◀Figure 1. the greater the impact of PDV on timing
1PPS Time Test 1PPS
Test platform for
Equipment
evaluating the impact of precision. Before transparent transport
DUT: Device under Test GPS: Global Positioning System Tod: Time of Day hops on time mode can be used to provide high
GE: Gigabit Ethernet PPS: Pulse Per Second synchronization. precision time synchronization signals,
much optimization must be done.
used in PTN, Packet Delay Variation random distribution.
(PDV) may have a significant impact on 1.3 Network Changeover
precision. Furthermore, network 1.2 PDV Changeover of time source, link, or
protection switching, signal PDV mainly impacts timing precision in clock board may lead to frequency
degradation, temperature, and the purely transparent transport mode. and/or phase shift, affecting timing
frequency synchronization may also As shown in Figure 3, all PTN nodes in precision. Test results show that time
impact timing precision. the test platform in Figure 1 are set to source changeover causes a timing
transparent transport mode. Timing difference of 6 ns, optical link
1.1 The Number of Hops difference is measured where there is changeover causes a difference of
In BC mode, each node terminates PTP no load, and a 90% load with packet 26 ns, and clock board changeover
packets, so PDV in the last node is not
accumulated. The main factor affecting 150
Timing Difference Range/ns
timing precision is the queue at the

100
output port. In addition, because slave
50
clocks at all levels are synchronized to
the master clock, drift is incurred during 0
clock recovery. This low-frequency -50
drift is accumulated. Figure 1 illustrates
-100
a test platform that is designed for
evaluating the impact of the number of -150
0 5􀆯
000 10􀆯
000 15􀆯 000 20􀆯000 25􀆯
000 30􀆯
000
hops on timing precision. On this Time/s
platform, all PTN nodes are configured (a) A Test with 10 Hops over 9 Hours
in BC mode and are interconnected by
250
Gigabit Ethernet (GE) interfaces. The
200
device under Test 1 (DUT1) is
connected to the GPS receiver with a 150
1 Pulse Per Second (1PPS) + Time of 100
Day (ToD) interface. Timing test 50
equipment is used to measure the 0
timing difference between the DUTn and -50
GPS receiver. -100
Figure 2 shows timing differences in 0 5􀆯
000 10􀆯
000 15􀆯000 20􀆯 000 25􀆯
000 30􀆯
000
Time/s
tests with different hops. In a nine hour
(b) A Test with 20 Hops over 9 Hours
test with 10 hops, the timing difference
50
ranges from -120.3 ns to 131.5 ns with
0
a peak-to-peak value of 252 ns. In a
nine hour test with 20 hops, the timing -50
difference ranges from -61 ns to 192 ns -100
with a peak-to-peak value of 253 ns. In -150
a four hour test with 30 hops, the timing -200
difference ranges from -239.3 ns to -250
26.8 ns with a peak-to-peak value of -300
266 ns. These test results show that 0 5􀆯
000 10􀆯
000 15􀆯
000
Time/s
timing precision changes only slightly in
(c) A Test with 30 Hops over 4 Hours
proportion to the number of hops, and
the noise model approximates ▲Figure 2. Timing difference with different hops.
Li Han
illustrated in Figure 5. Under normal

Peak-to-Peak Timing Difference/ns
250
conditions, the peak-to-peak timing

350
200
250 difference in tight coupling is 22 ns,
150
150
while that in loose coupling is 67 ns.
100 50
Synchronization in tight and loose
50 coupling is therefore similar.
-50
Figure 6 shows the impact of
0 -15
1 5􀆯
001 10􀆯
001 15􀆯001 20􀆯001 25􀆯
001 30􀆯
001 1 101201 301 401 frequency synchronization on timing
Time/s Time/s precision in tight coupling mode. When
(a) No Load (b) 90% Load with Packet Length of 64 bytes
the frequency is in Keep state, with an
2􀆯
000 Peak-to-Peak Timing Difference/ns 6􀆯
000
1􀆯
500 4􀆯
000 offset of 5×10-9, the peak-to-peak
1􀆯
000
2􀆯
000 timing difference is 100 ns. When the
500
0 0 frequency is in Free state, with an offset
-500 -2􀆯
000 of 3.8×10-8, time synchronization
-1􀆯
000
-4􀆯
000 signals are lost.
-1􀆯
500
-2􀆯
000 -6􀆯
000 The two coupling modes have their
11 101 201 301 401 1 201101 301 401
own advantages and disadvantages. In
Time/s Time/s
(c) 90% load with Packet Lenth of 576 Bytes (d) 90% load with Packet Length of 1􀆯518 Bytes the case of tight coupling, frequency
synchronization signals are degraded
▲Figure 3. Impact of PDV on transparent transport mode. or lost, which may affect time
synchronization signals. In the case of
causes a difference of 13 ns, as shown packets. Time synchronization in the loose coupling, each local PTN clock
in Figure 4. Generally, the timing cases of tight and loose coupling is may take a long time to trace the
difference arising from network
changeover is within 30 ns, enough for
5.84
a 50 ns allowance in the parameter
24.11
5.00
23.00
setting.
Timing Difference/ns
4.00
22.00 3.00
21.00 2.00
1.4 Signal Degradation and 20.00 1.00

19.00 0.00
Temperature Change 18.00 -1.26
Signal degradation may cause the rate 17.20
0 50 100 150 200 250 300 350 400 450 500 550 600 650 700 750 800 850 900 939
0 100 200 300 400 500 600
Time/s
700 800 900 1 000 1 060
of packet loss to increase, while Time/s
temperature change may affect clock

performance. In either case, timing (a) Phase/Time Change Arising from Time Source Changeover (b) Frequency Change Arising From Time Source Changeover
precision may be affected. When bits
are inserted with a Bit Error Rate (BER) 36.52 19.81
of 1× 10-3, the peak-to-peak timing 34.00 18.00

32.00 16.00
30.00 14.00
difference is within 110 ns, similar to the 28.00
26.00
12.00
case where there is no bit error. When 24.00 10.00

22.00 8.00
the temperature of PTN nodes is raised

20.00 6.00
18.00 4.00
16.00
from -10℃ to 50℃, the peak-to-peak 14.00 2.00
12.00 0.00
0 50 100 150 200 250 300 350 400 450 500 572
timing difference is within 40 ns, and
10.06 Time/s
0 50 100 150 200 250 300 350 400 450 500 550 577
Time/s
timing does not change proportionally
with temperature.
(c) Phase/Time Change Arising From Link Changeover (d) Frequency Change Arising From Link Changeover
1.5 Frequency Synchronization

PTN can obtain the clock from the 29.52 14.52
physical layer via synchronous

28.00
12.00
26.00
Ethernet, or recover the clock from

10.00
24.00
8.00
1588v2 packets. Tight coupling occurs 22.00
6.00
when PTN realizes frequency
20.00
4.00
18.00
synchronization using synchronous 16.00
15.26
2.00
0.26
Ethernet, and timing and phase 0 100 200 300 400 500
Time/s
600 700 800 900 10 00 10 48 0 50 100 150 200 250 300 350 400 450
Time/s
500 550 600 650 688
synchronization using 1588v2 packets.

Loose coupling occurs when PTN
realizes frequency, time, and phase (e) Phase/Time Change Arising From Clock Board Changeover (f) Frequency Change Arising From Clock Board Changeover
synchronization using only 1588v2 ▲Figure 4. Impacts of network changeovers on timing precision.
Special Topic Analysis of Time Synchronization in PTN
Li Han
1588v2 clock. Also, in loose coupling

mode, the frequencies of 1588v2
20.57 14.2
10.0
10.0
packets must be increased, which
18.00
0.0 increases the network load. Tight
16.00
14.00
12.00
-10.0 coupling is therefore preferred.
10.00 -20.0
8.00
6.00
4.00
-30.0
-40.0
2 Long-Term Stability of
Time Synchronization in
2.00
-0.86 -52.9
0 5 000 10 000 15 000 20 000 15 000 30 000 35 000 40 000 45 000 50 000 57.270 0 5 000 10 000 15 000 25 000 25 000 30 000 35 000 45 000 45 000 50 592
Time/s Time/s
BC Mode
To verify the stability of a large PTN
adopting 1588v2, part of the existing
network is selected for testing. The test
(a) Phase/Time Change in Tight Coupling (b) Phase/Time Change in Loose Coupling
network includes three core nodes,
▲Figure 5. Timing synchronization in tight and loose coupling. three aggregation nodes, and 58
access nodes. Each access node has
one TD-SCDMA base station, as
shown in Figure 7. All PTN nodes adopt
248.6 21 740
20 000
BC mode and all fiber segments
220.0 18 000 compensate for delay asymmetry.
Frequency synchronization is
200.0 16 000
180.0 14 000
160.0
12 000
10 000
implemented using synchronous
140.0
8 000 Ethernet. Time interfaces include
120.2
100.0
6 000
4 000
1PPS+ToD interface and FE/GE
80.0 2 000 interfaces.
66.1 -25
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 180 190 200 210 220
Time/s
0 50 100 150 200 250 300 350 400 450 500
Time/s
550 600 650 711 When FE is used, the peak-to-peak
timing difference over 72 hours is 65 ns,
as shown in Figure 8. The Maximum
Time Interval Error (MTIE) and Time
(a) Phase/Time Change with Frequency in Keep State (b) Phase/Time Change with Frequency in Free State Deviation (TDEV) results are shown in
▲Figure 6. Impact of frequency synchronization on timing precision in tight coupling mode. Figure 9. These results show that the
timing is stable, there is little noise, and
BITS_P BITS_W
the noise model approximates random
Core
distribution.
1 2 3 44
3 Conclusions
Aggregation Ring For TD-SCDMA and TD-LTE systems,
7 8
4 using GPS to implement time
24
5 6 25 synchronization is costly, insecure, and
difficult to engineer. Time
23 9 synchronization protocol has therefore
36
43
26
become the preferred technology for
35
22 transmitting high precision time
37 27 synchronization signals in fiber
21
42 10 systems. High precision time
34
38 synchronization in a PTN adopting IEEE
28 1588v2 has been proven feasible. In
20
41 40 39 11
existing networks, many fiber segments
are asymmetric, so compensation for
19
delay and jitter must be done segment
33 32 31 30 29
by segment. Both BC and TC eliminate
18 17 16 15 14 13 12
the impact of PDV, but BC is simpler.
BITS_P: Building Integrated Timing Supply_Protection BITS_W: Building Integrated Timing Supply_Working PTN: Packet Transport Network
Standards for time synchronization in
：PTN Type 1 ：PTN Type 2 ：PTN Type 3 ：10 GE ：GE ：1 588V2 & SyncE Path telecom networks are still immature,
and experience is limited. High
▲Figure 7. Network topology for timing synchronization test. precision time synchronization will first
Li Han
be deployed in PTN, and then in

Optical Transport Networks (OTN) and
42.3
Passive Optical Networks (PON) in the
future.
30.0
References
20.0
[1] Report of the Lannion SG15/13 interim meeting [R].
10.0 WD03, ITU-T SG15 Q13. 2009.
[2] LI Han, HAN Liuyan. Analysis of time
0.0 synchronization performance using PTP in hybrid
transport network [R]. CMCC, WD64, ITU-T SG15
-10.0 Q13. 2010.
[3] LI Han, HAN Liuyan. Proposal of accelerating MTIE
-21.9 and TDEV model study for time synchronization [R].
0 50 000 100 000 150 000 200 000 260 952 CMCC, WD63, ITU-T SG15 Q13. 2010.
Time/s [4] LI Han, WANG Lei. Test and analysis of time
synchronization using 1588v2 for transport network
[R]. CMCC, C599, ITU-T SG15 Q13. 2009.
◀Figure 8. [5] LI Han, WANG Lei. Time distribution model using
1588v2 for performance indication [R]. CMCC,
Long-term stability of time C601, ITU-T SG15 Q13. 2009.
synchronization.
Biography
1010 000 100.0
Li Han
1 000
10.0
Li Han received his Ph.D.
MTIE/ns
TDEV/ns
100
degree from Beijing University
10
1.0 of Posts and
Telecommunications. He is a
1 0.1 deputy principle at China
0.1 1.0 10.0 100.0 1 000.0 10 000.0 100 000.0 1 000 000.0 0.1 1.0 10.0 100.0 1 000.0 10 000.0 100 000.0 Mobile Research Institute,
Time/s Time/s
mainly engaged in the
research of transmission, wired
access and synchronization
technologies. He has
published over 50 papers in
(a) MTIE (b) TDEV
national and international journals, and applied for
MTIE: Maximum Time Interval Error TDEV: Time Deviation more than 20 patents. Being also an ITU-T editor,
he has delivered over 60 ITU-T contributions.
▲Figure 9. MTIE and TDEV.
Roundup
•• ••• • •• ••• ••• ••• • •• ••• • •• ••• ••• ••• • •• ••• • •• ••• ••• ••• • •• •••
•• ••• • •• ••• ••• ••• • •• ••• • •• ••• ••• ••• • •• ••• • •• ••• ••• ••• • •• •••
ZTE Positioned in the Leaders Quadrant in Leading Industry
•• ••• • •• ••• ••• ••• • •• ••• • •• ••• ••• ••• • •• ••• • •• ••• ••• ••• • •• •••
•• ••• • •• ••• ••• ••• • •• ••• • •• ••• ••• ••• • •• ••• • •• ••• ••• ••• • •• •••
Analyst Firm Report
•• ••• • •• ••• ••• ••• • •• ••• • •• ••• ••• ••• • •• ••• • •• ••• ••• ••• • •• •••
•• ••• • •• ••• ••• ••• • •• ••• • •• ••• ••• ••• • •• ••• • •• ••• ••• ••• • •• •••
ZTE Corporation announced on July best solution for every customer ZTE’s All-IP core network solution
14, 2010 that it has been positioned in requirement, and their products may covers fixed Softswitch, Mobile
the leaders quadrant in Gartner’s not be ‘best of breed’ throughout Softswitch, IMS and other fields, and
Magic Quadrant for Softswitch their portfolios. However, overall, they the continuous innovative Softswitch
Architecture Report. provide solutions that offer relatively and IMS solution promotes the healthy
As described in the report: "Leaders low risk and high quality." development of industry chain.
are high-viability vendors with a "We are honored to be one of the ZTE is becoming the leader in
broad portfolio, significant market Softswitch architecture leaders driving core network technology
share, broad geographic coverage, a recognized by Gartner’s Magic development and market application
clear vision for how service providers’ Quadrant Report. We believe that it is all over the world. Based on
needs will evolve, and a proven track a significant milestone for ZTE’s continuous investment in research and
record for delivering products. They sustainable developments and that it increasing in emerging markets, ZTE
are well positioned with their current is a clear endorsement of our gained 35% revenue growth in the
product portfolios and are likely to Softswitch and IMS solutions by the Softswitch market, while the overall
continue to deliver leading products. industry," said Mr. Xu Ziyang, Vice market suffered a decline of more than
Leaders do not necessarily offer the President of ZTE. 10%. (ZTE Corporation)
Research Papers Vehicular Ad-Hoc Networks: An Information-Centric Perspective
Bo Yu, Chengzhong Xu
Vehicular Ad-Hoc Networks: An

Information-Centric Perspective
Abstract:
Bo Yu
Emerging Vehicular Ad-Hoc Networks (VANET) have the potential to improve the
safety and efficiency of future highways. This paper reviews recent advances in
Chengzhong Xu
(Department of Electrical and Computer
wireless communication technologies with regard to their applications in vehicular Engineering, Wayne State University, Detroit,
environments. Four basic demands of future VANET applications are identified, and Michigan 48202, U.S.A.)
the research challenges in different protocol layers are summarized. Information
dissemination is one of the most important aspects of VANET research. This paper
also discusses the primary issues in information dissemination from an
information-centric perspective, and provides two case studies. Finally, future
research directions and possible starting points for new solutions are considered.
reports can then be delivered to a vehicles will also be equipped with an

1 Introduction user’ s cell phone via cellular networks. on-board computer, wireless radio,
V
ehicular Ad-Hoc Networks This is an expensive and inefficient way and a GPS receiver, which will enable
(VANET) are becoming an of disseminating location-based them to communicate with each other
integral technology for information, especially when the and with roadside units. A wireless
connecting daily life to computer information of interest is only a few sensor network of such magnitude is
networks. They could greatly improve hundred meters from the user’ s unprecedented, and perceptive
the driving experience both in terms of physical location. With its short-range computer systems will extend to every
safety and efficiency. As shown in communication capabilities, VANET corner of the globe. Information can be
Figure 1, when multi-hop may change this paradigm and make generated and shared locally in a
communication is implemented, VANET generating and disseminating peer-to-peer manner without the need
enables a vehicle to communicate with information more straightforward. for restrictive infrastructure.
other vehicles which are out of sight or VANET can also serve as a The capabilities of future vehicles
even out of radio transmission range. It large-scale wireless sensor network for open up a number of potential
also enables vehicles to communicate future ITS because every modern applications for use in daily life. The
with roadside infrastructure. VANET will vehicle can be regarded as a super main applications of VANET can be
likely be an essential part of future sensor node. For example, all new categorized as:
Intelligent Transportation Systems (ITS). vehicles are usually equipped with Safety applications: pre-collision
Currently, ITS relies heavily on exterior and interior thermometers, light warning, electronic road signs, traffic
infrastructure deployment. sensors, one or more cameras, light violation warning, online vehicle
Electromagnetic sensors, for example, microphones, ultrasound radar, and diagnosis, and road condition
are embedded into the road surface; other sensory features. Moreover, future detection. This type of application
traffic cameras are deployed at major
intersections; and Radio Frequency
Identification (RFID) readers are
deployed at highway entrances. A
typical procedure for collecting and
distributing traffic information is as
follows. First, traffic samples are
gathered by road surface sensors and
uploaded to a municipal transport Figure 1.▶
center. After data processing, traffic Vehicular ad-hoc networks.
Vehicular Ad-Hoc Networks: An Information-Centric Perspective Research Papers
usually takes advantage of Local businesses may gain a Interoperability for Microwave Access
short-range communication to perform competitive advantage or face greater (WiMAX) (IEEE 802.16). It is aimed at
real-time detection and provide competition. Undoubtedly, consumers providing wireless data over long
warnings to drivers will be the beneficiary of enhanced distances in a variety of ways, from
Efficiency applications: municipal safety and efficiency, cheaper goods, fixed point-to-point links to full mobile
traffic management, traffic congestion enriched entertainment, and other cellular type access. Currently, the
detection, route planning, highway advantages. most common form of automobile
tolling, and public transportation In this paper, the following section connectivity is based on cellular
management. This type of application is reviews recent advances in wireless telephony and is known as automotive
dedicated to improving both individual communication technologies with telematics. Typical examples include
and public travel efficiency regard to their applications in vehicular GM’ s OnStar system and Ford’s
Commercial applications: environments. Section 3 identifies four RESCU system. Several GIS
Location-Based Services (LBS) will fundamental demands of future VANET companies, including TomTom and
give rise to a variety of commercial applications. Section 4 discusses Garmin, also use cellular networks to
applications such as nearby restaurant existing challenges in different network transmit real-time traffic information.
specials, cheap gas stations, or even protocol layers. Section 5 further Usually, cellular-based telematics is a
shopping center promotions. Such discusses several research topics in paid service based on user
commercial applications may spur new information dissemination from an subscription.
competition among local businesses. information-centric perspective. In the near future, it is envisioned that
Infotainment applications: video and Section 6 concludes. architecture of vehicular networks will
music sharing, location-based be hybrid, as shown in Figure 2. In this
restaurant or store reviews, carpooling, 2 Wireless Technologies architecture, long-distance
and social networking. Already, communication techniques, such as
infotainment applications such as Ford
and Vehicular cellular networks and WiMAX, will
Sync[1] and Kia UVO have become Communications provide vehicles with instant Internet
attractive add-ons in the vehicle Wireless Personal Area Networks access, while short-distance
market. The networking of infotainment (WPAN) using IEEE 802.15 standards communication techniques, such as
systems will surely be a trend in the have been largely successful in Dedicated Short-Range
near future. consumer electronics (including Communications (DSRC)[6] and Wireless
An abundance of VANET vehicular electronics). Ford Sync is a Fidelity (Wi-Fi), will provide
applications will benefit a wide range of good example. With Bluetooth short-range real-time support in an ad
parties: from governments and vehicle technology, a cell phone can be hoc manner.
manufacturers to local retailers and connected to the vehicle’s audio VANET, based on DSRC, Wi-Fi, and
consumers. Although a few Geographic system enabling a driver to make calls other short range communication
Information Systems (GIS) or play hands-free music using voice techniques, will play an important role
companies—such as Google, Garmin, commands. 802.11 (a/b/g) WLAN in future ITS. Compared to
and TomTom—have engaged in technologies have been widely infrastructure networks, VANET has two
collecting and distributing traffic deployed because of their mass main advantages. It is cheap to deploy
information, traditionally, ITS production and relatively low cost. and operate, and consumers can enjoy
development and deployment has been Although 802.11 (a/b/g) was not service without subscription. VANET is
the domain of governments. In the originally designed for vehicular also essentially a cyber-physical
future, many more participants will be communications, many studies (in system, which enables communication
attracted to VANET and will profit from particular References[2-4]) have focused between two geographically
it. Vehicle manufacturers could predict on applying 802.11 to vehicular neighboring nodes. This has real-time
a boost in their sales by selling environments because of the safety and other applications.
VANET-enabled vehicles. Fitting pervasiveness of its technologies. IEEE
vehicles with a variety of electronic 802.11p[5] introduces enhancements to
controls and devices is a growing 802.11 which are needed to support
3 Requirements of VANET
trend, especially fitting electronic safety Wireless Access in Vehicular Applications
and information systems. Ford Sync is a Environments (WAVE). This includes Future VANET applications will have
very successful example of vehicle data exchange between high-speed four fundamental demands: scalability,
infotainment. Moreover, local retailers vehicles and between vehicles and availability, context-awareness, and
and service providers will also be roadside infrastructure in the licensed security and privacy.
interested in promoting their sales via ITS band of 5.9 GHz. (1) Scalability
VANET. They could broadcast Another emerging technology is Because of the number of vehicles
commercials to passing vehicles and Wireless Metropolitan Area Network that could be incorporated into
even devise hourly pricing strategies. (WirelessMAN), also called Worldwide vehicular networks, VANET may
reliability and scalability are two

DSRC: Dedicated Short-Range Communications requirements critical to safety
Internet
Wi-Fi: Wireless Fidelity applications. A number of traditional
WiMAX: Worldwide Interoperability for Microwave Access
link layer strategies, such as lengthy
access point selection, MAC
management timeout, and Address
Intra-Vehicle Resolution Protocol (ARP) timeout, have
Communications:
Bluetooth, Wi-Fi
Cellular Base Station been proven inefficient in highly mobile
3G, WiMAX
environments. They may lead to
increased start-up delays,
3G, WiMAX
underutilized bandwidth, or unfair
bandwidth allocation. Scalability and
reliability are interrelated issues.
DSRC Reliable broadcasting has been
intensively studied for vehicular safety
applications. The existing approaches
Internet DSRC include rebroadcasting, cooperative
forwarding, and transmission power
Roadside Unit
adaptation. However, reliability and
scalability remain open-ended issues
▲Figure 2. Wireless technologies for future vehicular communications. for safety applications because both
governments and consumers have
become the largest ad hoc network in The Ford Sync, for example, connects extremely high expectations for safety
history. Undoubtedly, scalability will be the vehicle’ s entertainment system to applications.
a critical factor. The advantages of the driver’ s cell phone via blue-tooth
hybrid architecture, together with technology. In the future, vehicular 4.2 Network Layer
in-network aggregation techniques on-board computers could even be In the network layer, the main challenge
and P2P technologies, make open to software developers. These is to establish a new paradigm for
information exchange more scalable. trends may have serious implications information dissemination in VANET.
(2) Availability for security and privacy due to the Ad hoc network routing has been
Due to the real-time interaction cyber physical nature of VANET. intensively studied in the last decade.
between vehicular networks and the Governments and consumers will have In particular, many context-aware
physical world, availability is an very high expectations of VANET safety routing protocols—such as
important factor in system design. This and security. Mobility-Centric Data Dissemination
may have a major impact on the safety Algorithm for Vehicular Networks
and efficiency of future highway (MDDV)[7] and Vehicle-Assisted Data
systems. The architecture should be 4 Research Challenges Delivery (VADD) in Vehicular Ad Hoc
robust enough to withstand unexpected This section discusses research Networks[8]—have been proposed for
system failures or deliberate attacks. challenges from different network VANET. These protocols significantly
(3) Context-Awareness protocol layers. The unique properties improve the packet forwarding
As a cyber-physical system, VANET of vehicular networks give rise to a performance in vehicular environments
collects information from the physical number of design challenges. These by taking advantage of vehicle mobility,
world and may conversely impact the properties also create new GPS position, and road layout. They are
physical world. On the one hand, opportunities to solve ITS problems essentially all packet-based; a packet
protocols should be adaptable to from a different perspective. travels from a source to a destination
real-time environmental changes, untouched throughout the entire
including vehicle density and 4.1 Link Layer process. However, this packet-based
movement, traffic flow, and road In the link layer, the main challenge lies paradigm no longer satisfies
topology changes. On the other hand, in adapting link layer protocols to application requirements in VANET
protocol designers should also unique vehicular environments and to from an information-centric
consider the possible consequences maximize link layer performance. There perspective. First, for some
the protocol may have on the physical are three major design objectives for applications, there is no definite source
world. link layer protocols: responsiveness, and destination, which is necessary for
(4) Security and Privacy reliability, and scalability. Link layer packet-based routing. Second,
There is a recent trend of making protocols are required to be highly information is altered (or combined)
vehicular on-board computer systems responsive to changes in channel throughout the forwarding process, and
inter-connectable to other systems. conditions and vehicle mobility, while this is not a consideration of packet
routing. In a traffic detection ▼Table 1. Representatives of information dissemination approaches

application, every vehicle may generate Information Dissemination Approaches Representative Work
a traffic report that can be combined
Infrastructure-Based Reference [9]
with other reports as it is disseminated.
For all interested vehicles intended to Macroscopic Delay Tolerant Routing References [10], [11], [12] and [13]
be the recipients of these reports, there Data Aggregation, Data Caching References [15], [18] and [24]
is no prior knowledge about how many, One-Hop References [2], [3] and [4]
when, or where these vehicles might
Microscopic Data Forwarding References [7], [8] and [20]
be. Some packet routing approaches
such as multicast and geocast can help Resource Management References [21], [22] and [23]
solve these issues. However, what is
needed is a new paradigm for a large-scale distributed database in macroscopic information dissemination
information routing—a replacement for which each vehicle maintains a local include information routing, data
packet routing. The new paradigm will part. Vehicles periodically exchange caching, and data aggregation.
enable information operations such as data to update this global database, Information dissemination can be
information generation, aggregation, and inconsistency cannot be avoided. implemented with or without
dissemination, and invalidation. Therefore, maintaining a relaxed infrastructure support. Jedrzej et al[9]
consistency model with minimal proposed establishing a P2P overlay
4.3 Application Layer overhead is a challenge. network based on cellular network
In the application layer, the challenge infrastructure. With a reliable
lies in effectively representing, connection to internet infrastructure,
discovering, storing, and updating 5 Information Dissemination vehicles can share, discover, and
information throughout the network. In this section, research demands and exchange information of non-safety
Naming and addressing are central challenges from an information-centric applications in a P2P manner. However,
problems in vehicular networks. How to perspective are discussed. VANET can service provided by infrastructure is
index information from the physical be regarded as an information-centric usually based on paid subscription,
world for efficient information storage system where information is collected which limits the number of consumers.
and dissemination remains an and disseminated throughout the Compared with infrastructure service,
unresolved problem. It is envisioned network, and it is important to identify ad hoc networks seem to be a more
that the addressing scheme will be a the system’ s demands from this attractive approach. On the other hand,
hybrid, multi-level scheme, with perspective. Information dissemination most non-safety applications do not
context information playing an can be classified into two levels: have strict real-time requirements. A
important role. The naming and macroscopic and microscopic. Two recent trend has been to study VANET
addressing policy has a significant case studies are presented for these information dissemination in a delay
impact on other system protocols, such levels respectively. Table 1 lists the tolerant manner. Some general purpose
as information discovery and routing. major research topics at these two DTN routing protocols such as
Because vehicles are highly mobile, levels and the representative work on epidemic routing[10] have been
another challenge is to dynamically each. proposed and evaluated. Proactive
map vehicle IDs to position-based approaches[11-12] take advantage of
addresses. This problem is particularly 5.1 Macroscopic Information apriori knowledge of geographical
important for applications across the Dissemination location, connectivity pattern, as well as
hybrid network architecture. Macroscopic information dissemination control over movement. Some existing
ARP/Reverse Address Resolution deals with disseminating information to DTN routing protocols assume a
Protocol (RARP)-like mechanisms can one node or a group of nodes in a predefined source and destination. For
be implemented in nodes equipped specified geographical area. The example, in Small and Hass’ s
with both DSRC and infrastructure destination of information can be a project[13], a DTN network was
network interfaces. single specified node in the network, a established for wild whale monitoring. A
Distributed data management is group of specified nodes, or even a sensor node was mounted on the back
another challenging issue for VANET, group of unknown nodes. The objective of a whale, and mobility information was
impacting data replication, data of macroscopic information delivered to an infostation hop by hop
elimination, and cache replacement. dissemination is to reduce information in a delay-tolerant manner.
Traditional distributed data delivery delay, reduce delivery Data caching and aggregation have
management assumes a network is overhead (including storage overhead also been studied in VANET.
connected with and communication overhead), and Zhao et al[14] studied the process of
geographically-distributed servers, increase the future query success rate distributing information from a data
which is no longer true for VANET. (if the destination is unknown in center into VANETs. Their approach
Essentially, VANET can be regarded as advance). General research topics for was based on periodical
rebroadcasting and buffering. It is a

one-way dissemination from a data
center to a group of vehicles. Lochert et Aggregation Area
al[15] proposed a hierarchical for Reports A & B
aggregation scheme that defines a set
of landmarks to calculate travel time.
They also proposed a roadside unit
placement algorithm to optimize A
aggregation.
B
Information dissemination, caching,
and aggregation have been individually
studied in relation to VANET. However,
the delay-tolerant dissemination : Dissemination Direction for Report A
problem has been fused into data : Dissemination Direction for Report B
query, data caching, and data
aggregation issues because, in VANET, ▲Figure 3. Dissemination tree.
for most types of information, there is no
a priori knowledge of the destination dissemination through an example of In order to aggregate the two reports,
vehicles. Any vehicle may generate and in-network data aggregation. As they must be delivered to the same
send out a query in the hope that a previously mentioned, in the near vehicle at the same time. For traditional
response is returned by a neighboring future, every vehicle will be a super wireless sensor networks, researchers
vehicle as soon as possible. sensor, capable of monitoring its have proposed a number of structural
A new paradigm for information surrounding environment. Each vehicle or semi-structural-based aggregation
routing needs to be established as an may generate a traffic report when the schemes involving the creation of a
alternative to packet routing. First, the vehicle speed is well below the speed transmission schedule to ensure
destination of information routing must limit. However, it is inefficient for every reports meet each other at the fork of a
be defined. The dissemination vehicle to generate a report and then routing tree. However, a fixed routing
destination is a virtual concept broadcast it to the entire network. It is structure is infeasable in VANETs.
constrained by time, space, and unnecessary to broadcast the speed of Several VANET projects, such as
vehicles. In other words, the destination individual vehicles on a road which is a Self-Organizing Traffic Information
consists of all vehicles which meet the few miles away; drivers expect to know System (SOTIS)[16] and TrafficView[17],
temporal and spatial conditions. There general congestion information ahead. use periodic rebroadcasting to collect
are two basic dissemination operations: Actually, traffic reports can be and redistribute traffic information.
pull and push. For pull, a vehicle combined as they are propagating, and Rebroadcasting is a feasible solution
periodically broadcasts its interest and the overhead for dissemination can be for local information exchange and
pulls data from other neighboring effectively reduced. dissemination, but it is difficult to scale
vehicles; for push, vehicles intentionally Figure 3 illustrates an example of it to city-wide dissemination.
push data to neighboring vehicles so report dissemination. Two vehicles A Essentially, in-network data
that other vehicles that may be and B, which are geographically close aggregation schemes trade off
interested in the data can easily obtain to each other, generate two traffic increased delay for reduced
it in the future. Since the pull operation reports. These two reports are intended redundancy. When a node receives a
is limited to one hop at the initial stage to be delivered to all vehicles following report, it introduces a delay before
of market penetration, it is more behind so that those vehicles can take forwarding it to the next hop so that it
important to devise push strategies. a detour before encountering a traffic may receive another report for this
When devising push strategies, the jam. A report will be duplicated at an duration. Structure-based schemes
potential impact to data caching and intersection, each copy going in one use a transmission schedule to
aggregation must be taken into direction. In this way, a dissemination determine this delay. Rebroadcasting
account. Heuristic neighbor information tree is formed. Since vehicles A and B schemes use a fixed delay for
(such as driving direction, speed, may not be within each other’ s signal rebroadcasting. If the delay is more
frequently visited places, etc.) or even range, one may not know that a similar adaptive, however, a packet is more
social networking information can be report was generated by the other. likely to meet other reports. In our
used to predict and control the Instead of being disseminated across previous work[18], intelligent delay
dissemination. the entire network, these two reports control policies based on local
can be merged as they propagate. In observations were investigated. For
5.1.1 In-Network Data Aggregation Figure 3, the dotted circle shows the example, suppose a node observes
This sub-section examines the details possible aggregation area for these two that a report has recently passed by; if
of macroscopic information reports. the node receives another report in a
Because of the short session duration

100
** : CATCHUP of a mobile-encounter scenario,
* efficient management of channel
80 * * : Randomized Waiting
resources is also a practical issue.
* Chang et al[21] proposed a scheduling
Number of Packets
60
* algorithm for downlinks—from a
** roadside unit to passing vehicles.
* ** ** Zhang et al[22] proposed another
20
**** scheduling algorithm which considers
* ***
* ** both upload requests and download
** ** *
20
**** requests. Yu et al[23] studied the
******* admission control problem when a
0
◀Figure 4. roadside unit is experiencing (or close
0 1 2 3 4 5 6 7 8 9 10
Performance of in-network to experiencing) overloaded conditions.
Distance/km
data aggregation. These studies improve the efficiency of
roadside unit access from different
short period, it can simply forward it to vehicles is important. resource allocation perspectives.
the next hop immediately in the hope A few recent research projects have In general, the main challenge in
that it can catch up with a previous one. paid close attention to one-hop microscopic information dissemination
If no report has recently passed, a long communication in a vehicular is how to bind lower layer conditions
delay can be applied in the hope that environment. Bychkovsky et al studied (mobility, channel, position) and upper
more reports can be later received by the techniques to increase one-hop layer application requirements together.
this node. A future reward model is throughput via open WiFi links. They From the upper layer perspective, DTN
designed to define the benefits of conducted a number of field tests to applications can tolerate information
different delay-control policies, and investigate the performance loss in delays and information inaccuracy;
then to establish a decision tree to help MAC association, IP address acquiring, from the lower layer perspective,
a vehicle choose an optimal policy from and IP route establishing. Hadaller et al mobility, channel, and location may
the perspective of long-term reward. conducted a 802.11-based one-hop change dramatically in a short period of
Figure 4 shows the performance of communication experiment and time. Existing work has studied the
such an aggregation scheme. Our furnished a detailed experimental one-hop communication problem with
scheme (CATCHUP) is compared with analysis. From their experiments, the different network protocols. However,
Randomized Waiting[19]; the results researchers identified the underlying there is still no effective bond between
show that the number of packets can causes of throughput loss in existing the upper and lower layers. The bond
be significantly reduced as the wireless access mechanisms. In sum, may allow us to take advantage of the
distance to the report source increases. these works attempt to analyze and lower harsh conditions, rather than
Data aggregation in VANETs has improve the link throughput from the being constrained by them.
attracted much research attention. perspective of lower layer protocols Three aspects are considered when
However, issues involving scalability, (phy, MAC, routing). designing microscopic information
data representation and processing, Microscopic information dissemination protocols.
and delay tolerant routing remain dissemination also deals with local (1) Application Requirement
unresolved. Among these, scalability is multi-hop communication. Usually, the DTN applications do not assume a
the most pressing issue. Although a main task of local multi-hop reliable link, but do prioritize data for
number of data aggregation schemes communication is to coordinate local transmission. They may specify the
have been proposed for VANETs, it is vehicles to disseminate information in a information loss tolerance level during
still not clear how scalable these predefined direction. VADD is a the dissemination process.
schemes are in terms of city-wide forwarding protocol which takes (2) Resource Management
communication. advantage of traffic pattern and road Problems include how to schedule
topology to source the best road for lower layer resources (such as
5.2 Microscopic Information delivering a packet. MDDV exploits transmission channel and transmission
Dissemination vehicle mobility for information rate); how to schedule upper layer
Microscopic information dissemination dissemination, and makes neighboring tasks; and how to allocate resources to
deals with information delivery in one vehicles collaborate in packet ensure fairness.
hop or in a few hops. In the initial forwarding in order to increase (3) Cooperation
market penetration stage, a vehicle reliability. Zhao et al[20] studied Here, focus is on cooperation
may rarely encounter another vehicle or throughput improvement gained between vehicles within signal range.
roadside unit. Therefore, increasing the through cooperative relaying to a Potential techniques include multi-task
efficiency of each encounter between roadside unit. scheduling, relaying, multi-party
the largest ad hoc network in history. In

20 the past few years, it has attracted
Failed Tasks (Percent)
* : RSUAC much attention from academia,

: Baseline
industry, and government. However,
20 there are fundamental issues that
remain unresolved. Better paradigms
are needed for information
0
4 *5 *6 *7 *8 *9 *
10 ◀Figure 5. dissemination and distributed data
The Number of Tasks for Each Vehicle Performance of admission management. Undoubtedly, the
control. number of research contributions will
continue to increase in the near future.
network coding, and others. admission control schemes mainly
focus on long-term sessions; for
5.2.1 Roadside Unit Admission Control example, VoIP and multimedia References
This subsection provides a case study services. Some real-time systems [1] Official ford sync site-Voice activated calling,
for microscopic information simply characterize transmission tasks navigation& music [EB/OL]. 2010-02-27. http://
www.fordvehicles.com/innovation/sync/.
dissemination. Roadside Units (RSU) by average rate, peak rate, or burst [2] OTT J, KUTSCHER D. Drive-thru Internet: IEEE
play an important role in future ITS. size. However, roadside unit access is 802.11b for "automobile" users [C]// Proceedings of
the 23rd Annual Joint Conference of the IEEE
They are usually deployed at highway mainly focused on short sessions, and Computer and Communications Societies
ramps or road intersections, and can the transmission rate for a moving (INFOCOM’ 04): Vol 1, Mar 7-11, 2004, Hong Kong,
communicate with passing vehicles vehicle may vary dramatically. A China. Piscataway, NJ, USA: IEEE, 2004: 362-373.
[3] BYCHKOVSKY V, HULL B, MIU A K, et al. A
with DSRC technologies. RSUs can dedicated admission control scheme is measurement study of vehicular Internet access
provide these vehicles with a variety of therefore desirable for roadside unit using in situ Wi-Fi networks [C]//Proceedings of the
12th Annual International Conference on Mobile
potential services. For example, a access. Computing and Networking (MobiCom’ 06), Sep
passing vehicle may download digital In our previous work, an admission 24-29, 2006, Los Angeles, CA, USA. New York, NY,
USA: ACM, 2006: 50-61.
maps, commercials, and traffic reports control scheme was proposed for
[4] HADALLER D, KESHAV S, BRECHT T, et al.
from an RSU. However, an RSU is a roadside unit access. The scheme Vehicular opportunistic communication under the
sparse resource. In the initial phase of calculates a transmission schedule for microscope [C]// Proceeding of the 5th International
Conference on Mobile Systems, Applications and
market penetration, only a limited all tasks (including current and new Services (MobiSys’ 07), Jun 11-13, 2007, San
number of RSUs can be deployed at tasks) based on a channel prediction Juan, Puerto Rico, USA. New York, NY, USA: ACM,
2007: 206-219.
highway ramps or major intersections. model and a vehicle mobility model. If a [5] IEEE 802.11p. IEEE draft standard for amendment
Even when a vehicle encounters a unit, feasible schedule is found, new tasks to standard [for] information
it may take less than half a minute to will be admitted; otherwise, new tasks technology-telecommunications and information
exchange between systems-local and metropolitan
move out of signal range. Moreover, will be rejected to guarantee the networks-specific requirements-Part II: wireless
multiple vehicles may concurrently success of current tasks. The problem LAN medium access control (MAC) and physical
layer (PHY) specifications-amendment 6: wireless
compete for services from the RSU. was treated as a linear-programming access in vehicular environments [S]. Jun, 2010.
Therefore, it is important to efficiently optimization problem and a set of [6] ARMSTRONG L. Dedicated short range
manage RSU access. algorithms were designed to calculate communications project [EB/OL]. http://www.
leearmstrong.com/DSRC/DSRCHomeset.htm.
Transmission integrity is also the bandwidth allocation schedule. All [7] WU Hao, FUJIMOTO R, GUENSLER R, et al. MDDV:
important for RSU access, since the concurrent transmission tasks share the a mobility-centric data dissemination algorithm for
vehicular networks [C]//Proceedings of the 1st ACM
services provided may be time or bandwidth according to the schedule, International Workshop on Vehicular Ad Hoc
location sensitive. If downloading a task thereby maximizing the success rate of Networks (VANET’04), Oct 1, 2004, Philadelphia,
like a digital map or traffic report cannot these tasks. In a NS2-based PA, USA. New York, NY, USA: ACM, 2004: 47-56.
[8] ZHAO Jing ,CAO Guohong. VADD:
be completed before the vehicle moves simulation, our scheme was compared vehicle-assisted data delivery in vehicular ad hoc
out of signal range, the downloaded with a baseline admission control networks [C]//Proceedings of 25th IEEE
International Conference on Computer
part would be meaningless, not to method. The baseline method allocated Communications (INFOCOM’ 06): Vol 5, Apr 23-29,
mention a waste of bandwidth. bandwidth based on a minimum 2006, Barcelona, Spain. Piscataway, NJ, USA:
Admission control is a potential required rate. Figure 5 demonstrates IEEE, 2006: 2528-2539.
[9] RYBICKI J, SCHEUERMANN B, KIESS W.
approach to guaranteeing transmission that the RSUAC scheme effectively Challenge: Peers on wheels-- a road to new traffic
integrity. The task of admission control reduced the percentage of failed tasks information systems [C]//Proceedings of the 13th
Annual International Conference on Mobile
is to determine whether to admit a new even when the workload (number of Computing and Networking (MobiCom '07), Sep
upload or download task. Once a task tasks per vehicle) increased. 9-14, 2007, Montreal, Canada. New York, NY, USA:
ACM, 2007: 215-221.
has been admitted, a transmission
[10] VAHDAT A, BECKER D. Epidemic routing for
schedule is calculated to guarantee partially-connected ad hoc networks [R].
completion of the task. 6 Conclusions CS-200006. Durham, NC, USA: Duke University,
2000.
Admission control has been the VANET is a promising area for future [11] YUAN Quan, CARDEI I,WU Jie. Predict and relay:
subject of intensive study. Traditional ITS, and has the potential to become an efficient routing in disruption-tolerant networks
[C]//Proceedings of 10th ACM International a data aggregation scheme for VANETs [C]// Jun 17-20, 2008, Beijing, China. Piscataway, NJ,
Symposium on Mobile Ad Hoc Networking and Proceedings of the 5th ACM International USA: IEEE, 2008: 731 - 739.
Computing (MobiHoc’ 09), May 18-21, 2009, New Workshop on Vehicular Ad Hoc Networks
Orleans, LA, USA. New York, NY, USA: ACM, (VANET’08), Sep 15, 2008, San Francisco, CA,
2009: 95-104. USA. New York, NY, USA: ACM, 2008: 49-57. Biographies
[12] GAO Wei, LI Qinghua, ZHAO Bo, et al. [19] FAN K W, LIU S, SINHA P. On the potential of Bo Yu
Multicasting in delay tolerant networks: a social structure-free data aggregation in sensor
network perspective [C]//Proceedings of 10th ACM networks [C]//Proceedings of 25th IEEE
International Symposium on Mobile Ad Hoc International Conference on Computer Bo Yu received his doctoral
Networking and Computing (MobiHoc’ 09), May Communications (INFOCOM’ 06): Vol 3, Apr degree from the department of
18-21, 2009, New Orleans, LA, USA. New York, 23-29, 2006, Barcelona, Spain. Piscataway, NJ, Computer Science, Fudan
NY, USA: ACM, 2009: 299-308. USA: IEEE , 2006: 1263-1274. University, China. Currently, he
[13] SMALL T, HAAS Z J. The shared wireless [20] ZHAO Jing, ARNOLD T, ZHANG Yang, et al. is a post-doc researcher in
infostation model: a new ad hoc networking Extending drive-thru data access by Wayne State University, USA.
paradigm (or where there is a whale, there is a vehicle-to-vehicle relay [C]//Proceedings of the His research interests include
way) [C]//Proceedings of 9th Annual International 5th ACM International Workshop on Vehicular Ad vehicular Ad hoc networks,
Conference on Mobile Computing and Networking Hoc Networks (VANET ’08), Sep 15, 2008, San wireless sensor networks, and
(MobiCom’03), Sep 14-19, 2003, San Diego, CA, Francisco, CA, USA. New York, NY, USA: ACM, mobile Ad hoc networks.
USA. New York, NY, USA: ACM, 2003: 233-244. 2008: 66-75.
[14] ZHAO Jing, ZHANG Yang, CAO Guohong. Data [21] CHANG C J,CHENG R G,SHIH H T, et al.
pouring and buffering on the road: a new data Maximum freedom last scheduling algorithm for Chengzhong Xu
dissemination paradigm for vehicular ad hoc downlinks of DSRC networks [J]. IEEE Transaction
networks [J]. IEEE Transactions on Vehicular on Intelligent Transportation Systems, 2007, 8(2):
Technology, 2007, 56(6): 3266-3277. 223-232. Chengzhong Xu received a
[15] LOCHERT C, SCHEUERMANN B, WEWETZER C, [22] ZHANG Yang, ZHAO Jing, CAO Guohong. On PhD degree from the University
et al. Data aggregation and roadside unit scheduling vehicle-roadside data access [C]// of Hongkong in 1993. He is a
placement for a VANET traffic information system Proceedings of the 4th ACM International professor of Electrical and
[C]//Proceedings of the 5th ACM International Workshop on Vehicular Ad Hoc Networks Computer Engineering and the
Workshop on Vehicular Ad Hoc Networks (VANET’07), Sep 10, 2007, Montréal, Canada. Director of Cloud and Internet
(VANET’08), Sep 15, 2008, San Francisco, CA, New York, NY, USA: ACM, 2007: 1-8. Computing Laboratory and
USA. New York, NY, USA: ACM, 2008: 58-65. [23] YU Bo, XU Chengzhong. Admission control for Sun’s Center of Excellence in
[16] WISCHHOF l, EBNER A, ROHLING H. Information roadside unit access in intelligent transportation Open Source Computing and
dissemination in self-organizing intervehicle systems [C]//Proceeding of 14th IEEE International Applications at Wayne State
networks [J]. IEEE Transactions on Intelligent Workshop on Quality of Service (IWQoS’ 09), Jul University. His research
Transportation Systems, 2005, 6(1): 90-101. 13-15, 2009, Charleston, SC, USA. Piscataway, interests include distributed and parallel computing
[17] NADEEM T, DASHTINEZHAD S, LIAO C, et al. NJ, USA: IEEE, 2009: 9p. and wireless embedded systems. He has published
Trafficview: Traffic data dissemination using [24] ZHAO Jing, ZHANG Ping, CAO Guohong. On two books and more than 150 papers and chaired
car-to-car communication [J]. Mobile Computing cooperative caching in wireless P2P networks [C]// numerous international conferences and workshops
and Communications Review, 2004, 8(3): 6-19. Proceedings of the 28th International Conference in these areas.
[18] YU Bo, GONG Jiayu, XU Chengzhong. Catch-up: on Distributed Computing Systems (ICDCS’ 08),
Roundup
•• ••• • •• ••• ••• ••• • •• ••• • •• ••• ••• ••• • •• ••• • •• ••• ••• ••• • •• •••
•• ••• • •• ••• ••• ••• • •• ••• • •• ••• ••• ••• • •• ••• • •• ••• ••• ••• • •• •••
ZTE and Three Release Android Éclair Handset in the UK
•• ••• • •• ••• ••• ••• • •• ••• • •• ••• ••• ••• • •• ••• • •• ••• ••• ••• • •• •••
•• ••• • •• ••• ••• ••• • •• ••• • •• ••• ••• ••• • •• ••• • •• ••• ••• ••• • •• •••
•• ••• • •• ••• ••• ••• • •• ••• • •• ••• ••• ••• • •• ••• • •• ••• ••• ••• • •• •••
•• ••• • •• ••• ••• ••• • •• ••• • •• ••• ••• ••• • •• ••• • •• ••• ••• ••• • •• •••
ZTE Corporation and Three have co-branded handset that combines a

launched the ZTE Racer, a low cost 2.8”QVGA touchscreen, 3.2
Android 2.1 (Éclair) powered handset. megapixel camera, 256 MB of internal
The Racer is the first Android memory, Bluetooth and 7.2 Mbit/s
handset in the UK from ZTE and is HSDPA capabilities.
Three’s first Android handset to break The handset is aimed at consumers
the 100 barrier, selling at 99.99. who require access to a wide range of
Android could account for the applications. The Android OS allows
second highest market share of users to download thousands of apps
mobile operating systems by 2012, from the Android market with quicklink
according to a recent report from access to services including
Gartner. However, it can take over a Facebook, Spotify, YouTube, Google
year to develop a new smartphone Maps and Google Talk.
which means that device ZTE expects to increase its share of
manufacturers are struggling to keep the UK carrier handset market from
up with developing Operating three percent in 2009 to eight percent
Systems (OS) and new devices are in 2010. In Q2 this year alone, ZTE
often launched with older versions of mobile devices sales in the UK
the OS. passed the million unit mark for the
The Racer is a ZTE and Three first time. (ZTE Corporation)
Research Papers Applying Analytical Hierarchy Process to a WiMAX Performance Evaluation Model
Ronald Shiu Keung Kwan, Kim Fung Tsang
Applying Analytical Hierarchy

Process to a WiMAX
Performance Evaluation Model
Abstract:
Ronald Shiu Keung Kwan1
Evaluating performance of individual features of WiMAX technology is a topic of
widespread discussion. Currently, there is no quantitative way of measuring WiMAX
Kim Fung Tsang 2
(1. Department of MEEM, City University of
technology so that wireless operators can meet their design objectives. This paper
Hong Kong, Hong Kong SAR, P. R. China;
outlines a set of design criteria for WiMAX and provides a decision-making aid that
2. Department of Electronic Engineering, City
ranks the importance of criteria using Analytic Hierarchy Process (AHP). This ranking University of Hong Kong, Hong Kong SAR, P.
should sufficiently reflect market expectations of the relative importance of various R. China)
design criteria. A model integrating AHP priorities with enhanced Data Envelopment
Analysis (DEA) is the basis for formulating a technological value in simple,
comparable format. A case study is provided to show how this technological value is
used to evaluate a three year network deployment plan. In the future, this model
could be extended to WiMAX equipment suppliers for the purpose of validating
performance targets of individual criteria, and enhancing supplier roadmaps for
future network development.
enhanced spectral efficiency, and to and Data Envelopment Analysis (DEA)

1 Introduction improve efficiency and coverage in can be used to develop a new WiMAX
Non-Line of Sight (NLOS) situations. performance model. The purpose of
T
o satisfy the need for multi-media A number of studies[1] have been such a model is to establish an
data access any time and carried out to evaluate different WiMAX agreed-upon performance index
anywhere, demand has grown for techniques, and to individually assess (defined as a technological value) that
high speed wireless data uplink/downlink data rate, spectral can be measured and compared with
transmission. Expansion of the smart efficiency, and Quality of Service (QoS). design objectives. It is built on the
phone and embedded device market However, such studies have failed to features of broadband wireless access
requires high speed data networks to reference overall design objectives. networks, but is not only applicable to
cope with different application suites. These assessments do not address the WiMAX technology. Indeed, the model
Worldwide Interoperability for fact that WiMAX technology should can easily be modified for Evolved High
Microwave Access (WiMAX) is based support multiple types of applications. Speed Packet Access (HSPA+) and
on the IEEE 802.16 standard. It was Thus, evaluation of any individual Long Term Evolution (LTE)
initially designed for fixed and nomadic feature does not help engineers technologies.
connection; and since 2001, has been improve design effectiveness in a
continually updated and enhanced. quantitative manner. 2 Methodology of
With the introduction of 802.16e in As a result, there is no consensus AHP-Based WiMAX
2005, it started to evolve towards about how good WiMAX technology
support for fixed and mobile and network performance can actually
Performance Evaluation
connection. Important technologies be. There is a lack of systematic and Model
such as Multiple-Input Multiple-Output analytical method to improve overall
(MIMO), Orthogonal Frequency Division network performance so that long term 2.1 AHP Modeling
Multiplexing (OFDMA), and Media high-speed data network development In the field of technology management,
Access Control (MAC) have been plans can be met. many studies[2] have been carried out in
introduced to enable high data rate, Analytical Hierarchy Process (AHP) which AHP has been adopted to
Applying Analytical Hierarchy Process to a WiMAX Performance Evaluation Model Research Papers
a maximum of 100%. This enables

technologies to be measured in a
Objectives(O ) Design Objectives simple and comparable format.
2.2 Integrating DEA into AHP Model

Criteria(Ck ) DEA is one of the key methods for
C1 C2 C3 Ck measuring technological efficiency of
WiMAX or Broadband Wireless Access
(BWA). It is a multi-factor efficiency
Sub-Criteria(Sj ) S1j S31 S32
analysis model for determining the
relative efficiencies of a set of decision
S13 making units or criteria.
S12 Technological efficiency is
S11 … S21 S22 S23 … S 2j Sk1 S k2 … Skj represented as a ratio of the maximum
sum of output to the relevant input
Technologies(Tn ) bundle. With this approach,
Technology
technological criteria are evaluated
Alternatives according to a semi-absolute value;
ideal technological efficiency in terms
▲Figure 1. An AHP hierarchical model using relative values. of network performance is 100.
Telecom systems—including wireless
evaluate or assess technology by They do not show how the technology networks—can be compared in this
prioritizing different criteria. AHP is a should progress in meeting objectives. way. However, never before has DEA
comprehensive framework designed to As a result, the conventional been applied to WiMAX or BWA
cope with intuitive, rational, and approach of constructing an AHP technology where multiple design
irrational factors when multi-objective, hierarchical model with technological criteria are used to assess efficiency.
multi-criterion decisions must be alternatives at the lowest level is Unlike traditional wireless
made, and a number of alternatives are insufficient for measuring overall communications in which capacity or
possible. WiMAX performance. A new spectral efficiency is the benchmark for
The AHP approach was designed to hierarchical approach called DEA[3] is assessing technology, multiple design
help decision makers integrate proposed in which the lowest level is criteria must be considered when
qualitative and quantitative aspects of replaced with technological efficiency, measuring the performance of WiMAX
complex problems. Complex problems as shown in Figure 2. Technological technology that supports not only voice,
are solved by decomposing the efficiency is a semi-absolute value with but also converged data, multi-media,
structure of the problem into
hierarchies, and pair-comparisons are
made to determine the priorities in each Objectives(O ) Design Objectives
hierarchy.
Typically, an hierarchical model
comprises three or four
levels—Objectives, Criteria, Criteria(Ck )
C1 C2 C3 Ck
Sub-Criteria, and Alternatives.
Objectives are on the top level while
alternatives are on the lowest level, as Sub-Criteria(Sj ) S1j S31 S32
shown in Figure 1. Pair-comparisons
determine the importance of design
S13
criteria to the design objective. The
S12
results are represented as a relative
S11 … S21 S22 S23 … S 2j Sk1 S k2 … Skj
value which indicates the technical
value of a criteria compared to others.
For example, technology 1 may be Technology E21,1 E2j,1 Ek1,1 Ekj,1
Efficiency (En )
twice as important as technology 2 in
… …
meeting design objectives. These E21,2 E2j,2 Ek1,2 Ekj,2
results are useful but insufficient
E21,n E2j,n Ek1,n Ekj,n
because they do not provide a holistic
measurement of the overall
performance of WiMAX technology. ▲Figure 2. A proposed hierarchical model using technological efficiency measurement.
WiMAX can be derived. This is a that BWA must overcome[4].

Objectives measure of WiMAX performance in (1) Data Rate
relation to technological expectations, Reliable transmission and reception
and can be represented by the schemes need to be developed to push
mathematical model: broadband data through hostile
Main Criteria
K J wireless channels. Typically, services
TVn= Σ Σ fk, j ×V(t +Wk×V(tn,k)
,k, j ) (1)
require high data rates and
n
k =1 J =1
In Equation (1), high-speed mobility. Wireless

Sub-Criteria •TVn: Technological value for communication relies on complex radio
measuring the performance of WiMAX wave propagation mechanisms so that
in relation to technological expectations; signals can traverse through the
Measure of Effectiveness •Wk: Relative priority of main intervening environment.
by Technology Walue criterion (k) with respect to the Several factors affect or impair
company’ s design objective; transmission:
▲Figure 3. An AHP hierarchical model using •fk, j : Relative priority of sub criterion •Decay of signal power in an NLOS
technological value for measurement. (j ) with respect to the main criteria (k); environment;
•tn,k, j: Performance and physical •Inter-symbol interference in a
and other IP applications. characteristics of design factor (n) multipath environment when time delay
By combining AHP and DEA along with main criteria (k) and sub between various signal paths becomes
methodology, design criteria can be criterion (j ); significant;
weighed and prioritized before being •tn,k: Performance and physical •Multipath fading caused by large
integrated into the DEA framework for characteristics of design factor (n) variations in the amplitude of the
deriving technological efficiency. The along with main criteria (k); received radio signals;
technological efficiencies for each •V(t ,k, j): Desirability value of the
n •Doppler effect due to relative
design criteria are then aggregated performance and physical motion between transmitter and
and transformed into the technology characteristics of technology (n) along receiver.
value, which is a quantifiable yardstick with sub criterion (j ) for main criterion (2) Spectral Efficiency
for measuring overall WiMAX (k); High spectral efficiency and broad
performance. •V(t ,k): Desirability value of the
n coverage must be achieved because
The methodology is as follows: performance and physical only a limited amount of spectrum is
(1) Performance measuring criteria characteristics of technology (n) for available to deliver broadband services
for WiMAX technology must be main criterion (k). to a large number of users. The
identified and categorized into main capacity of a wireless system is closely
performance groups. Main design related to its frequency usage. For
criteria and sub-criteria are used to
3 Characteristics of higher capacity and spectral efficiency,
build the AHP hierarchical model as WiMAX Technology frequency reuse must be maximized.
shown in Figure 3. WiMAX is a BWA technology, and BWA But this can significantly increase the
(2) Using AHP, the basic elements of is defined as wireless access with potential for interference. The
each level—such as the measuring broadband connection capabilities. challenge, therefore, is to design a
criteria and sub criteria—are then Broadband should have instantaneous transmission scheme that can operate
compared in pairs with respect to the bandwidths greater than 1 MHz and with a low Signal to Interference plus
design objectives. The comparison is support data rates greater than Noise Ratio (SINR).
based on a 9 point scale that indicates 1.5 Mbit/s (IEEE standard for Local and The signal level depends on cell
the relative importance of each Metropolitan area networks Part 16 radius and the number of base stations
criteria/sub-criteria. 2009). This is the bare minimum for within the coverage area. In order to
(3) DEA framework is used to widespread success. Broadband maintain a certain spectral efficiency
determine the technological efficiency wireless systems must deliver within the coverage area, the following
for each design criteria. These multi- megabit per second throughput should be considered:
technological efficiencies can then be to end users, and have robust QoS to •The cell radius of the base station;
used to evaluate the technology based support voice, data, and multimedia •The number of base stations and
on the degree to which its design services. Indeed, it should be able to cost required for the coverage area;
criteria satisfy the desirability on the support IP-based applications and •An effective method of ensuring
measure of effectiveness of each services in a highly efficient manner. reliability under low signal to SINR.
design factor. To meet stringent service (3) Policy Based QoS
(4) By integrating AHP priority with requirements and to overcome Broadband Wireless networks must
technological efficiency within the DEA constraints imposed by wireless, there support a variety of voice, data, and
framework, the technological value of are several technological challenges multimedia applications. Each
re-charging is not Although WiMAX employs a different

necessary for an extended RF access technology, integration of
period of time[7]. existing base stations, Access Services
To reduce power Network (ASN), and Connectivity
consumption, WiMAX Service Network (CSN) can bring about
technology should support seamless upgrade and easy migration
a power-efficient of the existing customer base[8].
transmission scheme,
power saving protocol for 4 AHP Hierarchical
fast switching technology,
signal processing algorithm,
Model for WiMAX
and it must be interoperable Performance Evaluation
with other radio channels AHP prioritizes and assesses network
(e.g. EVDO, Wi-Fi) within performance criteria/sub-criteria in
application has a different traffic pattern the subscriber station. relation to design objectives. It is
and QoS requirement[5]. Therefore, (6) Security particularly useful where certain design
policy-based QoS should be enforced Security is always a major criteria such as converged IP
so that service provision levels are consideration in wireless architecture and security are not
appropriate to different subscriber communication systems. End users are quantifiable. Paired comparisons
plans. often concerned about privacy and between criteria support logical and
QoS must be delivered end to end data integrity while network providers rational decision making.
across both the wireless link and are primarily concerned about AHP formulates a model by
underlying IP network. The challenge preventing unauthorized use of network structuring network performance
lies in accommodating variable QoS services. WiMAX networks should criteria in hierarchies consisting of three
requirements across application, support: levels: services operator’ s design
services, and user profile, while •Data encryption to avoid objectives at the highest level, followed
maintaining the quality levels as eavesdropping over the communication by categorized performance criteria,
defined for all users. Design factors link; and then performance sub-criteria at
affecting QoS include: •Authentication of legitimate end the lowest level.
•Throughput that handles multiple users who access the network;
traffic flow; •Access control over subscription
•Packet loss of different applications or services.
5 Case Study—An AHP
applications; (7) Converged IP-Based Architecture Model for Evaluating
•RAN latency in handover between IP-based architecture can facilitate Performace of an
cells; rapid development of new services and
•Jitter in the time between idle to applications because there is a large
Operator’ s WiMAX
transition. development community for leveraging. Network
(4) Mobility Management An IP-based system also tends to be The AHP model was adopted by a
Mobility is a key element for wireless cheaper because there are more leading WiMAX service provider to
services. However, support for a supplies in markets that have already measure their network performance
subscriber station moving over a large been adopted by wired communication against their desired target. The
coverage zone leads to two important systems. data—which was collected
design considerations[6], Although IP-based protocol is simple in 2009—included three years
•Roaming: A means of reaching and flexible, it is not necessarily (2009-2011) desired performance data
inactive users for session initiation and efficient and robust enough to operate for each design criteria/sub-criteria.
packet delivery must be provided with limited bandwidth in wireless The model converted performance data
regardless of their location within the environments. Thus, making IP protocol into technological efficiency, which
network; more bandwidth efficient and became an integral part of the
•Handoff: A continuous, guaranteeing QoS is of major concern technological value of the company’ s
uninterrupted session must be in WiMAX systems. WiMAX network. The technological
maintained with the base station while (8) Integration with Legacy value was an index for measuring the
on the move—even at vehicular speeds. Infrastructure extent to which the WiMAX network was
(5) Portability and Device For service operators with existing meeting the provider’ s desired
Management wireless infrastructure, it is important to performance level.
Portability requires the subscriber leverage the existing platform and roll
device to be lightweight, battery out wireless broadband services in a 5.1 Prioritization of the AHP Model
powered, and energy efficient so that timely and cost effective manner. Technological characterization was
▼Table 1. An AHP model with relative priority from a group of industrial experts defined by a group of WiMAX service
providers, who also provided
Relative Priority Relative Priority of
Main Criteria (Ck)
of main Criterion (Wk)
Sub-Criteria (Sj) Sub-Criterion (fk,j) judgement on the relative priority of
S11: Non Line of Sight Coverage design criteria/sub-criteria. An online
S12: Migitate Intersymbol Interference 0.12 evaluation tool was used during face to
C1: High Data Rate 0.312 S13: Diversity Management for Multipath 0.108
Fading 0.058 face interviews. Operators prioritized
S14: Suppression of Dopper Effect 0.026 the importance of each criteria by
C2: High Spectral
S21: Coverage by Increasing the Cell Radius
0.076 paired comparisons with other criteria
0.163 S22: Reliability Under Low Signal to SINR
Effciency
S23: Cost of Building Base Station
0.056 on the same level. The comparison
0.031
warranted the relative importance of
S31: Throughput Handling Multiple Traffic Flow
S32: Packet Loss for Support Various 0.036 each criteria and sub-criteria, and the
C3: Policy Based QoS 0.095 Applications 0.024 results are summarized in Table 1.
S33: Latency 0.025
S34: Jitter 0.01 The relative priority for the main
S41: Roaming criteria (Wk) and the sub-criteria (fk,j ) is
C4: Mobility Management 0.107 0.042
S42: Handoff Across Multiple Sites and Network
0.064 an integral part of the AHP model.
S51: Power Efficient Modulation to Support Low
Power Device 5.2 Technological Assessment with
0.046
C5: Portability and Device S52: Power Saving Protocol for Fast Switching
0.09 0.025 Desired Performance Targets
Management Algorithm
0.011
S53: Signal Processing Algorithm
S54: Interoperability with Other Radio Channel
0.009 The WiMAX operators estimated their
desired network performance levels
S61: Encryption to Protect Privacy
C6: Security 0.076 S62: Authentication for Identification
0.03 from 2009 to 2011, and this became the
0.033
S63: Access Control
0.013 desirable output under the DEA
framework. The data was then
C7: Low Cost Convergence S71: IP Protocol Bandwidth Efficiency
IP Architecture
0.127
S72: Delivery QoS Limited Bandwidth
0.04 integrated into the AHP model to
0.087
become the technological metrics for
S81: Integrate with Mobile Station
S82: Integrate at Access Service each of the measuring criteria, as
C8: Integration with Legacy 0.016
Infrastructure
0.031 Network (ASN)
0.008 shown in Table 2.
S83: Integrate at Connectivity Service
Network (CSN)
0.007 Desirability graphs representing
operators’preferences for the
▼Table 2. Technological metrics of the service provider’s WiMAX performance technological metrics of each design
criteria were drawn. Figure 4 shows
Main Criteria Sub-Criteria Measurement Unit Yr2009 Yr2010 Yr 2011
examples of desirability graphs for data
S11: Non Line of Sight Coverage UL/DL (Mbit/s) UL:4 UL:4.8 UL:5.2
S12: Migitate Intersymbol Interference DL:18 DL:21.6 DL:23.4 rate and spectral efficiency.
C1: High Data Rate
S13: Diversity Management for Multipath Fading The technological value represents
S14: Suppression of Dopper Effect
the overall performance of the AHP
S21: Coverage by Increasing the Cell Radius UL/DL (Mbit/s) UL:1.6 UL:1.92 UL:2.08
C2: High Spectral Efficiency S22: Reliability Under Low Signal to SINR DL:2.4 DL:2.88 DL:3.12 model after integrating all the
S23: Cost of Building Base Station technological efficiencies of the design
UL/DL (Mbit/s) UL:1.6 UL:1.92 UL:2.08 criteria. For the WiMAX study in 2009,
S31: Throughput Handling Multiple Traffic Flow
DL:2.4 DL:2.88 DL:3.12
S32: Packet Loss for Support Various Applications this was 50%. As WiMAX technology
C3: Policy Based QoS Percent <3% <2% <1%
S33: Latency
<ms 120ms 100ms 80ms continues to improve, the technological
S34: Jitter
<ms 80ms 60ms 40ms
AVe G VG
value is expected to improve to 70% by
Compare with 3G/HSDPA
(5-Very Good, 4-Good,
2010, and then to 100% by 2011.
S41: Roaming 3-Average, 2-Poor, 1-Very Poor)
C4: Mobility Management
S42: Handoff Across Mulitple Sites and Network 5 point scale AVe G VG
(5-Very Good, 4-Good,
3-Average, 2-Poor, 1-Very Poor)
5.3 Formation of the Performance
Evaluation Model
S51: Power Efficient Modulation to Support Low Power
Device
dB 10 20 20 The growth rate of technological value
mW 200 150 100
C5: Portability and Device S52: Power Saving Protocol for Fast Switching
mW 200 150 100 shows WiMAX service providers how
Management Algorithm
S53: Signal Processing Algorithm
No of different radio 0 1 2 design criteria should be improved in
channel (1 to 5)
S54: Interoperability with Other Radio Channel order to meet desired performance
S61: Encryption to Protect Privacy Compare with 3G/HSDPA VG VG VG targets.
C6: Security S62: Authentication for Identification (5-Very Good, 4-Good,
S63: Access Control 3-Average, 2-Poor, 1-Very Poor) High data rate (0.312) is still the
Relative cost compared with VG VG VG dominant factor affecting overall
C7: Low Cost Convergence S71: IP Protocol Bandwidth Efficiency 3G/HSDPA performance. Any improvement in
IP Architecture S72: Delivery QoS Limited Bandwidth (5-Very Good, 4-Good,
3-Average, 2-Poor, 1-Very Poor) NLOS coverage and/or interference
S81: Integrate with Mobile Station AVe G VG cancellation would also greatly improve
C8: Integration with Backward compatability
Legacy Infrastructure
S82: Integrate at Access Service Network (ASN) (5-Very Good, 4-Good, technological value. For example,
S83: Integrate at Connectivity Service Network (CSN) 3-Average, 2-Poor, 1-Very Poor)
technological value increases from 8%
Hierarchy Process (AHP) to build a strategic

100% ◆ framework for technology roadmapping [J].
Mathematical and Computer Modelling, 2007,
80% 46(7/8): 1071-1080.
[3] WANG Zhiqiang, ZHANG Lixin, QUE Huakun, et al.
Percentage
60% Evaluation of wireless access modes in distribution

◆
network communication based on weighted DEA
40%
[C]// Proceedings of the 1st International
◆
20% Conference on Sustainable Power Generation and
Supply (Supergen’ 09), Apr 6-7, 2009, Nanjing,
0% ◆
0 2 4 6 8 10
China. Los Alamitos, CA, USA: IEEE Computer
Mbit/s Society, 2009: 6p.
[4] ANDREWS J G, GHOSH A, MUHAMED R.
Data Rate Yr 2009 Yr 2010 Yr 2011 Fundamentals of WiMAX understanding broadband
UL (Mbit/s) 0.5 1 2 wireless networking [M]. Upper Saddle River, NJ,
USA: Prentice-Hall, 2007.
DL (Mbit/s) 2 4 8
[5] CHARILAS D, MARKAKI O, NIKITOPOULOS D, et
Total (Mbit/s) 2.5 5 10 al. Packet-switched network selection with the
(a) Desirability Graph for Data Rate and Capacity highest QoS in 4G networks [J]. Computer
Networks, 2008, 52 (1): 248-258.
[6] TUNG Hoi Yan, TSANG Kim Fung, LEE Lap To, et al.
100% ◆ On the handover performance of a tri-threshold
bandwidth reservation CAC scheme [J]. ETRI
80% Journal, 2007, 29(1): 113-115.
Percentage
60% [7] KIM Min Gon, CHOI Jung Yul, KANG Minko.
◆ Trade-off guidelines for power management
40% mechanism in the IEEE 802.16e MAC [J]. Computer
20% ◆ Communications, 2008, 31(10): 2063-2070.
[8] A comparative analysis of mobile WiMAX
0% ◆ deployment alternatives in the access network [S].
0 2 4 6 8 10 12
bit/s/Hz WiMAX Forum, 2007.
DL: Downlink Spectral Efficiency Yr 2009 Yr 2010 Yr 2011

UL: Uplink
UL (bit/s/Hz) 1 2 4
DL (bit/s/Hz) 2 4 8
Total (bit/s/Hz) 3 6 12
(b) Desirability Graph for Spectral Efficiency
Biographies
Ronald Shiu Keung Kwan
▲Figure 4. Desirability graphs for data rate and spectral efficiency.
Ronald Shiu Keung Kwan has
an MSc from City University,
to 16% between 2009 and 2010, and provides a quantitative approach to London, and an MEng. from
reaches 31% by the end of 2011. measuring an operator’ s WiMAX the University of Hong Kong.
He is a member of the British
High Spectral efficiency (0.163) is the network performance against their Computer Society, and the
second most important factor. Given desired targets. It is not designed for Institute of Electrical
that frequency spectrum is the most comparing one mobile operator’ s Engineering (IEE). He is
currently the assistant vice
limited resource in a wireless network network performance with another’ s, president of a global
ecosystem, operators are concerned as performance targets differ telecommunication company
and a doctoral student at City University, Hong
about enhancing efficiency and according to market strategy and Kong. He has over 20 years experience working in
maximizing the coverage and number investment model. the telecom industry in Hong Kong and North
of end users. Technological value for However, achieving targets very America. He has held several senior positions in
system and product development, solution
the operator increased from 4% to 8% much depends on equipment suppliers consulting, and network solution design. Kwan has
between 2009 and 2010, and will reach delivering products that align with the received a Premium Award for best paper from the
Institute of Electrical Engineering. His recent areas of
16% by 2011. operator’ s timetable. This model can research include Fixed Mobile Convergence (FMC),
The contribution of the above two also be extended to equipment wireless broadband access, and Ethernet backhaul.
criteria alone affects the overall WiMAX suppliers for the purpose of validating
network performance target by 47%. performance targets of individual Kim Fung Tsang
The other six design factors affect the criteria over some years of study, as
overall performance target by 53%. well as to enhance supplier roadmaps Kim Fung Tsang has a doctoral
degree from the University of
This model provides a clear indication for future network development. The Wales, Cardiff. He joined City
of what resource investments need to enhanced model could strengthen University of Hong Kong in
be made into the network, and monitors partnerships between mobile operators March 1988, and is now an
associate professor in the
performance improvement in a and equipment suppliers. Department of Electronic
quantitative way. Engineering. His current
research interests include radio
frequency circuit design,
References mobile communication network
6 Conclusions [1] WiMAX system evaluation methodology, V1.0 [S].
WiMAX Forum, 2007.
design, energy management, and wireless
automation protocol design.
The performance evaluation model [2] GERDSI N, KOCAOGLU D F. Applying the Analytic
Development Field Research on Convergence Network of EPON and WiMAX Based on ROF
Zhang Shu, Liu Deming, Wu Guangsheng
Research on Convergence Network

of EPON and WiMAX Based on ROF
and QoS performance[5]. However, the
Zhang Shu, Liu Deming, Wu Guangsheng BSs are required to convert digital
(College of Optoelectronic Science and Engineering, Huazhong University of Science and Technology,
baseband signals into wireless RF
Wuhan 430074, P. R. China)
signals, which may be very expensive
especially when wireless signals are
Abstract: high frequency. Large numbers of BSs
are required for wide coverage and to
compensate for the quick fading of
Convergence of optical and wireless networks is a promising development for future wireless signals.
access network architecture. A Radio over Fiber (ROF)-based network that converges The other solution is Radio over Fiber
Ethernet Passive Optical Network (EPON) and Worldwide Interoperability for Microwave (ROF) technology[6]. An ROF system
Access (WiMAX) technologies makes it possible to simultaneously transmit EPON uses optical fiber as the transmission
baseband signals and WiMAX wireless Radio Frequency (RF) signals. This article link between BS and Central Station
elaborates on uplink and downlink transmission, redundancy protection, and roaming (CS) and uses an optical carrier to
features of such a network. transmit RF signals. Optical fiber is only
used for transmission, while the CS is
responsible for switching, control and
signal regeneration. The BS performs
optical-electrical conversion. In this
way, complex and expensive devices
B
roadband access technology has Convergence of EPON and WiMAX can be placed at the CS to be shared
come into widespread use in line therefore avoids the drawbacks within by multiple remote BSs, and power
with wideband-consuming each individual technology, making full consumption costs can be reduced.
services such as Internet Protocol use of the wide bandwidth of optical Many technical papers have been
Television (IPTV) and Video on Demand access technology and flexibility of written on WLAN and ROF
(VoD). To further optimize resources, wireless technology. This gives users a convergence, as well as on successful
cut down costs, and provide wider and better experience, and also decreases commercial cases of GSM and ROF
more flexible broadband access construction and maintenance costs convergence. Research into
services, convergence of optical and across the entire network. transmission of WiMAX in Synchronous
wireless networks is possible in future Optical Networks (SONET) is currently
architecture[1-2]. At present, application 1 Current Solutions to being undertaken[7-8].
areas of Ethernet Passive Optical
Network (EPON)[3] are being expanded
EPON and WiMAX
2 Architecture of a
because of its low cost, wide bandwidth Convergence
and structural Ethernet advantages. On There are currently 2 solutions to EPON ROF-Based EPON and
the other hand, Worldwide and WiMAX convergence. One solution WiMAX Converged System
Interoperability for Microwave Access is baseband optical fiber transport In contrast to a standard EPON system,
(WiMAX) technology specified by IEEE technology, which transports digital a new Optical Line Terminal (OLT) and
802.16[4] is becoming a mainstream baseband signals directly through an Optical Network Unit Base Station
wireless broadband access optical fiber. The Optical Network Units (ONU-BS) are introduced in a
technology. Compared to Wireless (ONUs) of EPON and WiMAX Base ROF-based EPON and WiMAX
Local Area Network (WLAN) access Stations (BSs) are directly connected converged system.
technology, WiMAX provides wider through the Ethernet interface, and For the sake of reducing costs and
bandwidth, broader coverage and Ethernet frames are transmitted via concentrating processing, the devices
better Quality of Service (QoS). It also optical fiber. This scheme can be easily originally on the BS WiMAX physical
provides better data access service implemented, and has wide coverage. layers are moved to the central node
compared to cellular technology. It also has good bandwidth allocation OLT. The IEEE 802.16 physical layer
Research on Convergence Network of EPON and WiMAX Based on ROF Development Field
defines a working frequency of receives mixed signals from the OLT, it timeslots and their uplink RF signals do
2-66 GHz and 2 modulation modes: demultiplexes the baseband and not conflict with each other. Moreover,
single-carrier modulation and subcarrier signals, uploads the the radio RF signal has a frequency of
Orthogonal Frequency Division baseband signal to related devices for 3.5 GHz while that of the baseband
Multiplexing (OFDM) modulation. processing, and performs signal is below 2.5 GHz. Therefore, the
OFDM is used on 2-11 GHz frequency. down-conversion of the subcarrier baseband signal does not conflict with
Supporting Non-Line of Sight (NLOS) signal. The frequency of the local the radio RF signal, and they can be
transport, OFDM technology resists oscillation and the converter used is the transmitted at the same time. With this
attenuation and multipath. OFDM same as that of OLT. After moving design, the OLT receiver can be
modulation and demodulation devices through the band pass filter, the radio simplified by using only an
are used in this architecture. RF signal—originally belonging to the electrophotonic detector and a WiMAX
There are 3 usable frequencies under local BS—is demodulated and sent out OFDM demodulator. No local oscillation
11 GHz: 2.5 GHz, 3.5 GHz, and through the antenna. That is, the BS or converter is needed since there is no
5.8 GHz. Considering WiMAX only performs down conversion and cross-interference between
deployment, this architecture adopts requires no other devices. Costs are subcarriers.
3.5 GHz. Since there is no specifically therefore lowered.
defined carrier bandwidth, a WiMAX The ONU-BS uses Time Division 3 Features of the
system can work at a bandwidth Multiple Access (TDMA) mode to
between 1.25-20 MHz. This upload data in the uplink direction. In
ROF-Based Convergence
architecture adopts a bandwidth of this mode, the baseband and radio RF Solution
20 MHz, but ignores the problem of signals are mixed together. The EPON The architecture of this ROF-Based
co-channel interference. In actual MAC and WiMAX MAC cooperate with EPON and WiMAX converged system
application, frequency reuse and sector each other to allocate transmission has the following features:
partitioning technologies can make timeslots of uplink data for every (1) The BS performs down conversion
better use of frequency resources and ONU-BS. When a designated timeslot only. Devices responsible for radio
system throughput. In the deployment appears, the ONU-BS sends data signal processing and some
of an ROF system, direct modulation is according to the authorized OLT upper-layer functions are placed at the
usually adopted if the RF signals are window size. In this way, uplink data of OLT end. For ease of description, the
below 10 GHz, and external modulation all ONU-BSs can be transported in a OLT end shown in Figure 1 uses N
is above 10 GHz. Direct modulation is pre-defined sequence without conflict OFDM modulators (one for each BS),
used in this architecture. after they have reached the shared which in practise can be combined into
In the downlink direction, this optical fiber. Performing up-conversion one. At the OLT end, all BSs can share
architecture adopts Subcarrier on the radio RF signal to convert it into a one set of OFDM modulation and
Multiplexing (SCM) technology at the subcarrier signal at the ONU-BS end is demodulation devices.
OLT end in order to simultaneously unnecessary. Different ONU-BSs are (2) N+1 (rather than N+N ) protection
transmit EPON wireline baseband located in different transmission is implemented because all devices are
signals and wireless RF signals, and
also to distinguish RF signals that
belong to different BSs. This OLT OLT ONU-BS1
UP
ONU
DW
UP DW
architecture defines the EPON 20 MHz
baseband signals at 0-2.5 GHz, and BS1 3.5 GHz 3.5 GHz
up-converts the radio signals to OFDM
Modulator
BPF
Wireless Subcarrier BPF
subcarriers above 3.5 GHz. The LO1 LO1
bandwidth of each subcarrier is Σ 3.5 GHz Σ
…
WiMAX MAC
20 MHz, and the central frequency Σ Σ

interval is 0.1 GHz. One subcarrier BS n 3.5 GHz 2.5 GHz 3.6 GHZ
OFDM
corresponds to one BS; for an EPON Modulator BPF
system with a branching ratio of 1:16, LOn

LOn-1+0.1 GHz ONU-BS2
there are 16 subcarriers in total. The
laser is modulated after the baseband OFDM
…
Demodulator 20 km
signal and modulated subcarrier have ONU-BSn
been integrated. Since the baseband
BPF: Band Pass Filter OLT: Optical Line Terminal
signal and subcarrier remain at different BS: Base Station OFDM: Orthogonal Frequency Division Multiplexing
frequencies, no interference is EPON: Ethernet Passive Optical Network ONU: Optical Network Unit
LO: Local Oscillator ROF: RF Optical Fiber Transmission
produced. Baseband and radio signals MAC: Media Access Control WiMAX: Worldwide Interoperability for Microwave Access
are simultaneously transported in the
downlink. When the remote ONU-BS ▲Figure 1. ROF-based EPON and WiMAX converged system architecture.
Development Field Research on Convergence Network of EPON and WiMAX Based on ROF
moved to the OLT end. N BSs share just modulation and detection technology, Proceedings of the 1st International Conference on
Communications and Electronics (ICCE’ 06), Oct
one spare BS. In current radio especially when the radio frequency is 10-11, 2006, Hanoi, Vietnam. Piscataway, NJ, USA:
networks, BSs cannot support high. Moreover, as it takes time for the IEEE, 2006.
redundancy protection, or can only radio signal to be transported over
provide 1+1 protection. In other words, optical fiber, the longer transmission
one spare BS is needed for every BS. In distance impacts MAC performance[9] Biographies
an ROF-based converged network, more significantly. Further research is Zhang Shu
system stability is improved and cost is required into the bandwidth allocation
reduced. algorithm so that uplink transmission Zhang Shu received his
doctoral degree from the
(3) OLT serves as a central controller timeslots can be better allocated for College of Optoelectronics
that manages the wireless resources of wireline baseband data and radio RF Science and Engineering,
Huazhong University of
all BSs and knows the real-time radio signals. Science and Technology. His
resource information of all BSs. In research interests include
standard WiMAX architecture, wireless optical fiber communications,
optical fiber access networks,
resource management is performed at 4 Conclusion fiber-wireless convergence,
different BSs and information cannot be Architecture that converges optical and home network
technologies. He has participated in numerous
exchanged within the necessary fiber and wireless networks has a bright projects funded by the National High Technology
timeframe. This limits the system’s work future. An ROF-based EPON and Research and Development Program of China ("863"
program), China Hubei Provincial Science &
efficiency. With the help of an algorithm, WiMAX converged network Technology 10th 5-year program, and National
OLT dynamically allocates wireless simultaneously transmits the EPON Defense Pre-Research Foundation Program of
resources for every BS in order to make baseband signal and WiMAX radio RF China. He has published more than 10 papers and
applied for 3 national invention patents.
full use of the wireless resource and to signal via optical fiber. Such a system
balance the loads among BSs. System combines fixed line and mobile
efficiency is accordingly improved, and networks, and brings about more Liu Deming
can be further improved if satisfactory user experience. Moreover,
Liu Deming is a professor and
Multiple-Input and Multiple-Output it can greatly reduce network doctoral advisor at the College
(MIMO) and adaptive modulation construction and maintenance costs. of Optoelectronics Science and
Engineering, Huazhong
technologies are employed. University of Science and
(4) Roaming among BSs is much References
Technology. He is also the
director of the National
simpler. OLT processes all data and [1] SUEMURA Y, NAKAMURA S, HU J. Converged
Engineering Laboratory for
knows all information, so the control optical/wireless systems architecture [C]//
Internet Access System. His
Proceedings of the Asia-Pacific Optical
channel between OLTs and every BS is Communications (APOC’ 06), Sep 5, 2006,
research work covers optical
fiber communications
no longer necessary. To support user Gwangju, South Korea. SPIE 6354.
technology, optical fiber sensing technology, optical
[2] SHEN G X, TUCKER R S, CHAE C J. Fixed and
mobility, roaming in different WiMAX mobile convergence architectures for broadband
network technology, solar energy photovoltaic
technology, and semiconductor lighting technology.
BSs must be taken into consideration. access: integration of EPON and WiMAX [J]. IEEE
He has won the third prize in the National Invention
BSs cannot connect with each other, Communications Magazine, 2007, 6(8): 44-50.
Awards, the first prize in the National Natural Science
[3] IEEE 802.3ah. Amendment: media access control
and in any ordinary networking scheme, parameters, physical layers, and management
Awards, and the second prize in the Hubei Provincial
Technological Invention Awards. He has published
a control channel is used—between parameters for subscriber access networks [S].
more than 100 papers and applied for more than 30
2004.
OLT and every WiMAX BS—to [4] IEEE 802.16e. Air interface for fixed and mobile
national invention patents.
exchange control information in real broadband wireless access systems [S] 2006.
time. When a user roams from one BS [5] 常宇光, 刘德明, 张曙, 等. 基于固网移动网融合(FMC)
的宽带接入网系统 [J]. 光通信技术, 2009, 33(5):
Wu Guangsheng
to another, OLT sends a command via 45-49.
this control channel to make the original CHANG Yuguang, LIU Deming, ZHANG Shu, et al. Wu Guangsheng received his
Fixed Mobile Convergence (FMC) for broadband undergraduate degree from
BS disconnect the user and at the same access system [J]. Optical Communication Tianjin University and his
time add them to the local BS. However, Technology, 2009, 33(5): 45-49. master’s and doctoral degrees
[6] 曹培炎. ROF 技术在无线接入网络中的应用 [J]. 光通 from the College of
in converged architecture, when OLT
信技术, 2005, 29(10): 22-26. Optoelectronics Science and
senses that a user is roaming between CAO Peiyan. The implementation of the ROF Engineering, Huazhong
different BSs, it simply switches over to technology in the wireless access network [J]. University of Science and
Optical Communication Technology, 2005, 29(10): Technology. His research
a new subcarrier frequency to send 22-26. interests include optical fiber
data to the user. [7] SARKAR S, DIXIT S, MUKHERJEE B. Hybrid access network technology,
Wireless-Optical Broadband-Access Network fiber-wireless convergence technology, and next
The conveged system, however, is (WOBAN): a review of relevant challenges [J]. generation optical network technology. He has
not without some weaknesses. It is Journal of Lightwave Technology, 2007, 25(2): participated in numerous projects funded by the
susceptible to fiber attenuation, 3329-3340. National High Technology Research and
[8] GHAZISAIDI N, PAOLUCCI F, MAIER M. Development Program of China ("863" program), and
dispersion, and non-linear effect in SuperMAN: Optical-wireless integration of RPR and China Hubei Provincial Science & Technology 10th
optical fiber, as well as cross WiMAX [J]. Optical Networking, 2009, 18 (3): 5-year program. He has published more than 20
249-270. papers, and applied for 10 national invention
interference between subcarriers. It [9] DANG B L, PRASAD V, NIEMEGEERS I. On the patents.
also places high requirements on MAC protocols for radio over fiber networks [C]//
Cloud Computing (3) Lecture Series
Wang Bai, Xu Liutong
Cloud Computing (3)

services and solutions vendors,
partners, and end users that provide or
Wang Bai consume shared resources within the
cloud computing environment.
Xu Liutong Collaboration between vendors and
their partners is emphasized in the
(School of Computer Science and Technology, Beijing University of Posts and Telecommunications, cloud computing value chain.
(2) Virtualization for Cloud
Infrastructure
Hardware virtualization involves
Editor's Desk: managing hardware equipment in
In the preceding two parts of this series, several aspects of cloud computing — including plug-and-play mode; software
definition, classification, characteristics, typical applications, and service levels — were virtualization involves using software
discussed. This part continues with a discussion of Cloud Computing Oopen Architecture image management or code
and Market-Oriented Cloud. A comparison is made between cloud computing and other
virtualization technology to enable
software sharing. Dynamic code
distributed computing technologies, and Google’s cloud platform is analyzed to
assembly and execution is another
determine how distributed computing is implemented in its particular model.
software virtualization technology. In an
Internet application, some JavaScript
code elements can be dynamically
retrieved and inserted into an Ajax
package to create new functions or
6.4 Open Architecture Such an ecosystem includes all features for a web client.
V
irtualization is a core technology
for enabling cloud resource Cloud Ecosystem
1
sharing, and Service-Oriented Cloud Client Dashboard
Architecture (SOA) enables (Portal or Program-Based Access Interactions)
1D
flexibility, scalability, and reusability. By Cloud Offerings: Business Solutions
1C
(Unified Representation of Cloud Resources & Exchanged Message)
combining these two technologies, 5

Cloud Value Added Services or Offerings
researchers have developed an open
(QoS, Monitoring, Metering Billing, Exception Handling)
(Infrastructure as a service, Software as a Service, Application as a Service, Business Process as a Service)

cloud computing architecture based (Onboarding and Membership Management)
on the Open System Interconnection

Cloud Quality and Governance
Cloud Ecosystem Management
Cloud Core: Provisioning and Subscription Services

Cloud Information Architecture
(Portal or Program-Based Access Interactions)
(OSI) model, and have used it as the Cloud Provisioning Service Cloud Subscription Service
4
reference model for implementing the
Cloud Partner Dashboard
(Provisioning Process, Roles, Notification) (Subscription Process, Roles, Notification)

cloud computing system, as shown in
Figure 6. Service-Orientation: Common Reusable Service & Composite Servcies
This architecture encompasses cloud 3

Cloud Horizontal Business Cloud Vertical Business
ecosystem, cloud infrastructure and its Services (CRM, ERP) Services (Industry-Specific Services)
management, service-orientation, core
Virtualization: Hardware and Software
provisioning and subscription,
2
composite cloud offerings, cloud Cloud IT Infrastructure Management
information architecture and (Management of Hardware, Software, and Legacy Application)
Data
management, and cloud quality Cloud Core Infrastructure Center
Servers
analytics. In designing the architecture, Mainframe (Hardware, Software, and Legacy Applications)
6 7 1A
seven basic principles are adopted:
Cloud Vendor Dashboard
(1) Integrated Management for Cloud 1B (Portal or Program-Based Access Interactions)
Ecosystem
CRM: Customer Relationship Management ERP: Enterprise Resource Planning QoS: Quality of Service
An architecture must support cloud
computing ecosystem management. ▲Figure 6. Cloud computing open architecture.
Lecture Series Cloud Computing (3)
(3) Service-Orientation
Users/Brokers
Service-orientation is a driving force
that gives cloud computing business
value in terms of asset reusability, Service Request Examiner and
Admission Control
composite applications, and mashup •Customer-Driven Service Management
•Computational Risk Management
services. Common services can be •Autonomic Resource Management
SLA Resource
reused to enable the cloud’s core Allocator
provisioning and subscription services Pricing Accounting
as well as to build cloud offerings in

VM Service Request
Infrastructure as a Service (IaaS), Monitor Dispatcher Monitor
Software as a Service (SaaS), and even
Business Process as a Service (BPaaS). VMs
(4) Extensible Service Provisioning
This feature is unique to cloud
computing systems. Without Physical Machines
extensibility, the provisioning part of the
cloud architecture can only support a
certain type of resource sharing. Free
use and paying users can periodically SLA: Service Level Agreement
VM: Virtual Machine
change their roles as service providers
or consumers and this change can ▲Figure 7. High-level market-oriented cloud architecture.
occur at three levels of service
provisioning. quality indicators for the cloud Agreement (SLA)-based resource
(5) Configurable Enablement for computing environment and a set of allocation in the context of cloud
Cloud Offerings guidelines to govern the design, computing, as shown in Figure 7.
The architecture must ensure deployment, operation, and In this architecture, there are four
configurability of the cloud computing management of cloud offerings. main entities:
platform and services. The modularized In short, the objective of such an (1) Users/Brokers
ecosystem management, virtualization, architecture is to combine SOA and Users (or brokers acting on their
service-orientation, and cloud core virtualization technologies in order to behalf) submit service requests from
form a solid foundation to ensure a exploit the business potential of cloud anywhere in the world to the Cloud
computing platform that is configurable, computing. Computing Center to be processed.
combinable and manageable. (2) SLA Resource Allocator
(6) Unified Information 6.5 Market-Oriented Cloud The SLA Resource Allocator acts as
Representation and Exchange Cloud computing is a new the interface between Cloud service
Framework Internet-based resource sharing mode provider and external users/brokers. It
The collaborative feature of cloud particularly focused on its business requires the interaction of the following
computing comprises information model. How, then, does this feature mechanisms to support SLA-oriented
representation and message exchange impact cloud computing? Researchers resource management.
between cloud computing resources. have proposed a market-oriented •Service Request Examiner and
Cloud computing resources include all cloud architecture, global cloud Admission Control
business entities (e.g. cloud clients, exchange and market infrastructure for When a service request is first
partners, and vendors) and supporting trading services, which have been submitted, the Service Request
resources such as virtualization related investigated intensively. Examiner and Admission Control
modules, service-orientation related mechanism interprets it for QoS
modules, cloud core, and cloud 6.5.1 Market-Oriented Cloud requirements before determining
offerings. The cloud information Architecture whether to accept or reject it. The
architecture module enables In the article Cloud Computing and mechanism also requires updated
representation of cloud entities in a Emerging IT Platform: Vision, Hype, and status information on resource
unified cloud computing entity Reality for Delivering Computing as the availability from the Virtual Machine
description framework. Message 5th Utility, researchers from the Cloud (VM) Monitor mechanism and workload
routing and exchange protocols as well Computing and Distributed Systems processing from the Service Request
as message transformation capability (CLOUDS) Laboratory of the University Monitor mechanism in order to make
form the foundation of cloud information of Melbourne presented a effective resource allocation decisions.
architecture. market-oriented architecture. This It then assigns the request to a VM and
(7) Cloud Quality and Governance architecture supports Quality of Service determines resource entitlements for
This module identifies and defines (QoS) negotiation and Service Level the allocated VM.
tools, and technologies must be

Compute Cloud developed to represent, convert, and
enhance resource value. Figure 8
Storage Cloud illustrates a cloud exchange and
Broker 1 market system model based on
Requet Negotiate/Bid
Capacity Publish real-world exchanges.
Offers In this model, the market directory
Directory allows participants to locate providers
or consumers with suitable offers.
Bank
Auctioneers periodically clear bids and
…
Enterprise Auctioneer requests received from market

Resource
Manager
participants, and the banking system
Broker N
(Proxy) Compute carries out financial transactions.
Global Cloud Cloud
Market Brokers perform the same function in
Enterprise IT Consumer such a market as they do in real-world
markets: they mediate between
Storage Cloud
consumers and providers by
▲Figure 8. Global cloud exchange and market infrastructure for trading services. purchasing from the provider and
sub-leasing to the consumer.
•Pricing encompass both customer-driven Consumers, brokers and providers are
The Pricing mechanism determines service management and bound to their requirements and related
how service requests will be charged computational risk management in compensations through SLAs. An SLA
based on submission time order to sustain SLA-oriented resource specifies the details of the service to be
(peak/off-peak), pricing rates, or allocation; provided in terms of metrics agreed
resource availability. •Incorporate autonomic resource upon by all parties, and penalties for
•Accounting management models that effectively violating these expectations,
The Accounting mechanism meters self-manage changes in service respectively. Such markets can bridge
the actual usage of resources by each requirements in order to satisfy both disparate clouds, allowing consumers
request so that the final cost can be new service demands and existing to choose a suitable provider by either
calculated and charged to the user. service obligations; executing SLAs in advance or by
•VM Monitor •Leverage VM technology to purchasing capacity on the spot.
The VM Monitor mechanism oversees dynamically assign resource shares Providers can set the prices for a
the availability of VMs and their according to service requirements. resource based on market conditions,
resource entitlements. user demand, or current level of
•Service Request Monitor 6.5.2 Cloud Service Exchanges and utilization of the resource. The
The Service Request Monitor Markets admission-control mechanism at the
mechanism oversees the execution Enterprises currently employ cloud provider end is responsible for
progress of service requests. services to improve the scalability of selecting the auctions to participate in
(3) VMs their services and to deal with bursts in or the brokers to negotiate with. The
Multiple VMs can be activated or resource demand. However, at present, negotiation process continues until an
stopped dynamically on a single the proprietary interfaces and pricing SLA is formed or the participants
physical machine to meet accepted strategies of service providers prevent decide to break off. Brokers profit from
service requests. consumers from swapping one provider the difference between the cost of
(4) Physical Machines for another. For cloud computing to leasing the resource, and what they
Multiple computing servers form a become mature, services must follow charge consumers to gain a share of
resource cluster to meet service standard interfaces. This would enable the resource. A broker, therefore, must
demands. services to be commoditized and would choose both consumers and providers.
Commercial market-oriented cloud pave the way for the creation of a Consumer demands include
systems must be able to: market infrastructure for trading in deadlines, fidelity of results, turnaround
•Support customer-driven service services. time of applications, and budget
management; In cloud computing markets, service limitations. Enterprise consumers can
•Define computational risk consumers expect their specific QoS deploy their own limited IT resources
management tactics to identify, assess, requirements to be met with minimal into clouds as guarantees for
and manage risks involved in the expense, and service providers hope to enterprise computing, or they can lease
execution of applications; retain their clients while achieving the providers’resources to upscale their
•Devise appropriate market-based highest possible Return on Investment applications.
resource management strategies that (ROI). To achieve this, mechanisms, The idea of utility markets for
computing resources has been around ▼Table 1. Key characteristics of clusters, grids, and cloud systems
for a long time. Recent research Systems
projects have particularly focused on Characteristics
Clusters Grids Clouds
trading VM-based resource allocation High-End Computers Commodity Computers, High-End
Population Commodity Computers
by time slices. In the above model, a (Servers, Clusters) Servers, and network Attached Storage
resource broker can negotiate with Size/Scalability 100 1,000 100 to 1,000
resource providers. Based on Node OS Linux, Windows Mainly Unix

Multiple OSs Running on Virtual
Machines
enterprise Grid, Melbourne University’s Ownership Single Multiple Single
CLOUDS Laboratory implements a Dedicated, High-End Network
Interconnection Mostly Internet with High Dedicated, High-End Network with
market-oriented platform called Network/Speed
with Low Latency and High
Latency and Low Bandwidth Low Latency and High Bandwidth
Bandwidth
"Aneka", which is also a NET-based
Traditional Public/Private Key Pair Based Each User/Application is Provided with
service-oriented resource Security/Privacy
Login/Password-Based; Authentication and Mapping a Virtual Machine; High
management platform. Aneka exhibits Medium Level of Privacy
(Depends on User Privileges).
a User to an Account; Limited Security/Privacy is Guaranteed; Support
Support for Privacy. for Setting Per-File ACL.
many of the properties of the cloud Centralized Indexing and
computing model. Discovery Membership Services Decentralized Information Membership Services
Services
Service Negotiation Limited Yes, SLA Based Yes, SLA Based
6.6 Comparison of Cluster, Grid, and
User Management Centralized Decentralized, VO-Based Centralized or Delegated to Third Party
Cloud Computing
Resource Management Centralized Distributed Centralized/Distributed
The first part in this series briefly
Allocation/Scheduling Centralized Decentralized Centralized/Decentralized
introduced some characteristics of Some Open Grid Forum
cloud computing that can be directly Standards/Inter-Operability VIA-Based
Standards
Web Services (SOAP and REST)
experienced by users. This part, Single System Image Yes No Yes, but Optional
however, discusses some of the Capacity Stable and Guaranteed Varies, but High Provisioned on Demand
technical characteristics that Failure Management
Limited (Failed Limited (Failed Strong Support for Failover and Content
Tasks/Applications are Tasks/Applications are Replication; VMs can be Easily
distinguish cloud computing platforms (Self-Healing)
Generally Restarted). Generally Restarted) Migrated from One Node to Another
from cluster and grid computing. Service Pricing Limited, in Non-Open Market Mainly Internal Pricing
Pricing According to Effects, with
Discounts for VIP Clients
Although cloud platforms share some
Highly Potential; Support Third-Party
common characteristics with clusters Internetworking
Multi-Clustering within an
Organization
Limited Solution Providers’Loose Coupling of
and Grids, they have their own unique Services from Different Clouds
Collaborative Scientific and
attributes and capabilities. These Application Driven
Science, Business, Enterprise
Computing, Data Centers
High Throughput Computing
Dynamically Provisioned Legacy and
Web Applications, Content Delivery
include support for virtualization, Applications
Highly Potential; Create New Services
services with Web Service interfaces Potential for Building by Dynamically Provisioning of
Limited, Mainly Scientific
that can be dynamically composed, Third-Party Value-Added
Solutions
Limited
Computing Oriented
Computing, Storage, and Application
Services and Offer Individual or
and support for the creation of Combined Cloud Services to Users
third-party, value-added services by ACL: Access Control List SOAP: Simple Object Access Protocol VO: Virtual Organization
building on cloud compute, storage, OS: Operating System VIA: Virtual Interface Architecture
REST: Representational State Transfer VM: Virtual Machine
and application services. Table 1
compares the key characteristics of
cluster, grid and cloud computing computers, creating a powerful data but generate new computing data.
systems. data center. Therefore, one feature of GFS is that it
generates a large number of very large
7.1 GFS File System files mainly for reading. These files can
7 Cloud Computing Model Desktop applications differ from be appended but rarely re-written. GFS
Although enterprises and academic Internet applications in many respects. is also characterized by high data
researchers have proposed various GFS is a proprietary distributed file throughput.
cloud system models, most of these do system developed by Google Inc. It is There are two types of GFS nodes:
not reveal the computing paradigm for designed to allow efficient and reliable one master node, and a large number
problem solving in a could. To enable access to data by using large clusters of chunkservers. Chunkservers store
communication and collaboration of commodity hardware. GFS is data files, with each individual file
between server clusters within a cloud, optimized for Google’ s core data broken up into fixed-sized chunks of
Google has introduced Google File storage and usage needs (primarily the 64 megabytes. Each chunk is assigned
System (GFS), BigTable, and search engine), which can generate a unique 64-bit label to maintain logical
MapReduce technologies—so-called enormous amounts of data that needs mappings of files to constituent chunks.
the "three sharp weapons" of cloud to be retained. Google’ s Internet The master node only stores metadata
computing. With these technologies, search computing learns from the associated with the chunks, such as the
Google has formed a cloud with functional programming paradigm in tables mapping the 64-bit labels to
thousands or even millions of which operations do not modify original chunk locations and the files they make
each word in a large set of documents.

In functional programming, map and
reduce are tools for constructing
higher-order functions.
Map applies a given function to a list
of elements (element by element) and
returns a new list. These new elements
are the products of the function applied
to each element in the original list. For
example, Map f [v1, v2, ..., vn ] = [f (v1),
f (v2), ..., f (vn)]. In this way, the
functions can be computed in parallel.
The MapReduce computing model is
up, the locations of the copies of the data layout, storage format, and suitable for applications requiring
chunks, what processes are reading or storage location, BigTable meets high-performance parallel computing.
writing to a particular chunk, or taking a application demands for localized If the same computing is required on a
"snapshot" of the chunk pursuant to access. Tables are optimized for GFS, large set of data, the data set can be
replicating it. This metadata is kept being split into multiple tablets of about divided and assigned to different
current by the master node as it 200 megabytes. The locations in the machines for computing.
periodically receives updates from GFS of tablets are recorded as Reduce involves combining elements
each chunk server. database entries in multiple special of a list using a computing approach
Modification permissions are handled tablets, which are called "META1" (function). To unfold a binary operation
by means of time-limited "leases". The tablets. META1 tablets are found by (function) into a n-ary operation
master node grants permission for a querying the single "META0" tablet. The (function), the reduce function is used:
process to modify a chunk within a META0 tablet typically has a machine to Reduce f [v1, v2, ..., vn ]=f (v1, (Reduce
given period. The modifying itself which is queried by clients by f[v2, ..., vn]) = f (v1, f (v2, (Reduce f [v3, ...,
chunkserver, which is always the clients for the location of the META1 vn])) = f (v1, f (v2, f ( ...f (vn-1, vn) ...)).
primary chunk holder, then propagates tablets; and consequently, the location MapReduce computing model
the changes to chunkservers with of the actual data. combines the intermediate results
backup copies for synchronization. BigTable is designed for databases obtained from Map operations by
With several redundant copies, of petabyte scale with data across Reduce operations until the final result
reliability and availability are thousands of servers. It is also is calculated.
guaranteed. Programs access the designed to accomodate more
chunks by first querying the Master machines without the need for 7.3.2 Execution Procedure
server for the locations of the desired reconfiguration. Map invocations are distributed across
chunks, the Master replies with the multiple machines by automatically
locations, and the program then 7.3 MapReduce Distributed partitioning the input data into a set of
contacts and receives the data from the Programming Paradigm splits or shards. Reduce invocations
chunkserver directly. GFS and BigTable are used by Google are distributed by partitioning the
Google currently has over 200 GFS for reliable storage of data in a intermediate key space into pieces
clusters, each of which consists of large-scale distributed environment. using a partitioning function. When the
1,000 to 5,000 servers. Using GFS, Google’ s MapReduce is a software user program calls the MapReduce
Google has proven that clouds built on framework designed to support parallel function, the overall operation flow is
cheap machines can also deliver computing on large data sets (often illustrated in Figure 9.
reliable computing and storage. greater than 1 terabyte) on a large The MapReduce library in the user
cluster. It is therefore a computing program first splits the input files into M
7.2 BigTable Database System model specifically designed for cloud pieces. It then starts up many copies of
BigTable is a compressed, high computing. the program on a cluster of machines.
performance proprietary database One of the copies of the
system built mainly on GFS and 7.3.1 Software Framework program—the master—is special; the
Chubby Lock Service. It is also a The MapReduce software framework rest are workers. The master picks idle
distributed system for storing structured design is inspired by two common workers and assigns each one a map
data. A BigTable is a sparse, programming functions: "Map" and task or a reduce task.
distributed, multi-dimensional sorted "Reduce". It was developed within A worker that is assigned a Map task
map, which can be indexed by a row Google as a mechanism for processing reads the contents of the
key, column key, and a timestamp. By large amounts of raw data; for example, corresponding input split. It parses
allowing a client to dynamically control counting the number of occurrences of key/value pairs out of the input data and
MapReduce and GFS, Hadoop is an

User open-source Java software framework
Program
(1) Fork (1) Fork
consisting of functional
programming-based concurrent
(1) Fork
computing model, and a distributed file
Master system. Hadoop’ s HBase, similar to
(2) Assign (2) Assign BigTable distributed database,
Map Reduce supports data-intensive distributed
applications to work with thousands of
Worker nodes and petabytes of data.
Hadoop was originally developed to
Split 0 (6) Write Output
(5) Remote Worker File 0
support distribution for the Nutch
Split 1 Read
(3) Read
(4) Local search engine project. Yahoo has
Write
Split 2 Worker invested a great deal of money into the
Split 3 Output
project and uses Hadoop extensively in
Split 4
Worker File 1 areas such as web search and
advertising. IBM and Google have
Worker
launched an initiative to use Hadoop to
support university courses in
Input Map Intermediate Files Reduce Output
Files Phase (on Local Disks) Phase Files distributed computer programming. All
these have been instrumental in
▲Figure 9. Google MapReduce execution procedure. promoting and popularizing cloud
computing worldwide.
passes each pair to the user-defined reduce function. The output of the (To be continued)
map function. The intermediate reduce function is appended to a final
key/value pairs produced by the Map output file for this reduce partition. Biographies
function are buffered in memory. When all Map and Reduce tasks Wang Bai
The buffered pairs are periodically have been completed, the master
written to local disk and partitioned into wakes up the user program. At this Wang Bai is a professor and
vice dean of the School of
R regions by the partitioning function. point, the MapReduce call in the user Computer Science and
The locations of these buffered pairs on program returns back to the user code. Technology, Beijing University
of Posts and
the local disk are passed back to the Upon completion, the output of the Telecommunications (BUPT).
master, which is responsible for MapReduce execution is available in Her research interests include
forwarding these locations to the the R output files. Typically, users do next generation telecom
operation supporting system,
Reduce workers. not need to combine these R output distributed computing
When a Reduce worker is notified by files into one file; they often pass these technologies, and visualization
analysis of complex networks. She has published
the master about these locations, it files as input to another MapReduce more than 60 papers and 3 books.
uses remote procedure calls to read the call or use them from another
buffered data from the local disks of distributed application.
map workers. When a Reduce worker Xu Liutong
has read all intermediate data for its 7.4 Apache Hadoop Distributed System
partition, it sorts it by the intermediate Infrastructure Xu Liutong is a professor at the
School of Computer Science
keys so that all occurrences of the Google’ s GFS, BigTable, and and Technology, BUPT. His
same key are grouped together. MapReduce technologies are open to research interests include data
The Reduce worker iterates over the the public but their implementation is mining, grid computing, and
distributed systems and their
sorted intermediate data and for each private. A typical implementation of application in telecom field. He
unique intermediate key encountered, it these technologies in the open source has published more than 20
papers.
passes the key and the corresponding community involves the Apache
set of intermediate values to the user’s Hadoop project. Inspired by Google’ s
••••••••••••••••••••••••••••••••••••••••••
••••••••••••••••••••••••••••••••••••••••••
•••••••••••••••••••••••••••••••••••••••••• A1,Back Cover:
AD Index ••••••••••••••••••••••••••••••••••••••••••
••••••••••••••••••••••••••••••••••••••••••
•••••••••••••••••••••••••••••••••••••••••• ZTE Corporation
••••••••••••••••••••••••••••••••••••••••••
Abbreviation Index
Abbreviation Index
A CSN: Connectivity Service Network H
ACL: Access Control List HDLC: High Level Link Control
AF: Assured Forwarding D HSDPA: High Speed Downlink Packet
AGW: Access Gateway DEA: Data Envelopment Analysis Access
AHP: Analytical Hierarchy Process DF: Default Forwarding HSPA: High Speed Packet Access
APS: Automatic Protection Switching DFS: Depth-First Search
ARP: Address Resolution Protocol DL: Downlink I
ASN: Access Services Network DSCP: Differentiated Services Code IaaS: Infrastructure as a Service
ATM: Asynchronous Transfer Mode Point IMA: Inverse Multiplexing over ATM
DSRC: Dedicated Short-Range IPTV: Internet Protocol Television
B Communications IS-IS-TE: Intermediate System to
BA: Behavior Aggregates DUT: Device Under Test Intermediate System
BC: Boundary Clock Extensions for Traffic
BER: Bit Error Rate E Engineering
BITS_P: Building Integrated Timing E2E: Edge-to-Edge ITS: Intelligent Transportation Systems
Supply_Protection EBS: Excess Burst Size
BITS_W: Building Integrated Timing ECMP: Equal-Cost Multi-Path Routing L
Supply_Working EDGE: Enhanced Data rates for GSM LACP: Link Aggregation Control
BMCA: Best Master Clock Algorithm Evolution Protocol
BPaaS: Business Process as a Service EF: Expedited Forwarding LAG: Link Aggregation
BPF: Band Pass Filter EIR: Excess Information Rate LAN: Local Area Network
BRAS: Broadband Remote Access EPON: Ethernet Passive Optical LBS: Location-Based Service
Server Network LMP: Link Management Protocol
BS: Base Station ERP: Enterprise Resource Planning LSP: Label Switched Path
BSC: Base Station Controller ESMC: Ethernet Synchronous Message LTE: Long Term Evolution
BTS: Base Transceiver Station Channel
BWA: Broadband Wireless Access ESP: Ethernet Switched Path M
MAC: Media Access Control
C F MAN: Metropolitan Area Network
CBR: Constant Bit Rate FEC: Forwarding Equivalence Class MBMS: Multimedia Broadcast and
CBS: Committed Burst Size FR: Frame Relay Multicast Service
CCAMP: Common Control and FSM: Finite State Machine MC-HSPA: Multi-Carrier HSPA
Measurement Plane FTP: File Transfer Protocol MDDV: Mobility-Centric Data
CDR: Call Detailed Record Dissemination Algorithm for
CE: Customer Edge G Vehicular Networks
CEP: Circuit Emulation over Packet GE: Gigabit Ethernet MIMO: Multiple-Input Multiple-Output
CESoPSN: Circuit Emulation Services GFP: Generic Framing Procedure MLPPP: Multi-Level Pre-Emptive
over Packet Switch Network GFP-F: Generic Frame Procedure Priority
CFM: Connectivity Fault Management Framed MMS: Multimedia Messaging Service
CII: Common Interworking Indicator GFS: Google File System MP2MP: Multipoint-to-Multipoint
CIR: Committed Information Rate GIS: Geographic Information System MPLS: Multiprotocol Label Switching
CO-CS: Connetion-Oriented GMPLS: Generalized Multiprotocol MPLS-TE: Multiprotocol Label
Circuit-Switched Label Switching Switching Traffic Engineering
CRM: Customer Relationship GPRS: General Packet Radio Service MPLS-TP: Multiprotocol Label
Management GPS: Global Positioning System Switching Transport Profile
CS: Central Station GUI: Graphical User Interface MS-PW: Multi-Segment Pseudo Wires
Abbreviation Index
MSTP: Multiservice Transport Platform PSN: Packet Switched Network TDMA: Time Division Multiple Access
MTIE: Maximum Time Interval Error PTN: Packet Transport Network TD-SCDMA: Time Division
PTP: Precision Time Protocol Synchronous Code Division
N PW: Pseudo Wire Multiple Access
NE: Network Element PWE3: Pseudo Wire Emulation TE: Traffic Engineering
NLOS: Non-Line of Sight Edge-to-Edge T-LSP: Transport Label Switched Path
NTP: Network Time Protocol PWES: Pseudo Wire Emulation Service TLV: Type-Lengh-Values
T-MPLS: Transport Multiprotocol Label
O R Switching
OAM: Operation, Administration and RAN: Radio Access Network ToP: Time over Packet
Maintenance RARP: Reverse Address Resolution TPS: Tributary Protection Switching
OC: Ordinary Clock Protocol TTL: Time-to-Live
ODU: Optical Cannel Data Unit REST: Representational State Transfer
OFDM: Orthogonal Frequency Division RFID: Radio Frequency Identification U
Multiplexing RNC: Radio Network Controller UDP: User Datagram Protocol
OLT: Optical Line Terminal ROF: Radio over Fiber UL: Uplink
ONU: Optical Network Unit RPR: Resilient Packet Ring
OS: Operating System V
OSI: Open System Interconnection S VADD: Vehicle-Assisted Data Delivery
OSPF-TE: Open Shortest Path First with SaaS: Software as a Service VANET: Vehicular Ad-Hoc Network
Traffic Engineering SAToP: Structure-Agnostic TDM over VBR: Variable Bit Rate
Extensions Packet VC: Virtual Channel
OTM: Optical Transport Module SCM: Subcarrier Multiplexing VCC: Virtual Channel Connection
OTN: Optical Transport Network SD: Signal Degrade VIA: Virtual Interface Architecture
SDH: Synchronous Digital Hierarchy VLAN: Virtual Local Area Network
P SF: Signal Failure VM: Virtual Machine
P2MP: Point-to-Multipoint SINR: Signal to Interference plus Noise VO: Virtual Organization
P2P: Peer-to-Peer Ratio VoD: Video on Demand
PBB: Provider Backbone Bridge SLA: Service Level Agreement VoIP: Voice over Internet Protocol
PBBN: Provider Backbone Bridge SMS: Short Message Service VP: Virtual Path
Network SNCP: Subnetwork Connection VPC: Virtual Path Connection
PBB-TE: Provider Backbone Bridge Protection VPN: Virtual Private Network
Traffic Engineering SOA: Service-Oriented Architecture VS: Virtual Section
PBT: Provider Backbone Transport SOAP: Simple Object Access Protocol
PDH: Plesiochronous Digital Hierarchy SONET: Synchronous Optical Network W
PDV: Packet Delay Variation SOTIS: Self-Organizing Traffic WAVE: Wireless Access in Vehicular
PE: Provider Edge Information System Environments
PHB: Per-Hop Behavior SR: Service Router WCDMA: Wideband Code Division
PHP: Penultimate Hop Popping SSM: Synchronization Status Message Multiple Access
PIR: Peak Information Rate SS-PW: Single-Segment Pseudo Wire WDM: Wavelength Division Multiplexing
PLL: Phase-Locked Loop STM: Synchronous Transport Module Wi-Fi: Wireless Fidelity
PLSB: Provider Link State Bridging WiMAX: Worldwide Interoperability for
PON: Passive Optical Network T Microwave Access
PPP: Point-to-Point Protocol TC: Transparent Clock WLAN: Wireless Local Area Network
PPS: Pulse Per Second TDEV: Time Deviation WPAN: Wireless Personal Area Network
PRC: Primary Reference Clock TDM: Time Division Multiplexing WTR: Wait to Restore

Ontents: SPECIAL TOPIC: Packet Transport Networks

Diunggah oleh

Informasi Dokumen

Judul Asli

Hak Cipta

Format Tersedia

Bagikan dokumen Ini

Bagikan atau Tanam Dokumen

Opsi Berbagi

Apakah menurut Anda dokumen ini bermanfaat?

Apakah konten ini tidak pantas?

Hak Cipta:

Format Tersedia

Ontents: SPECIAL TOPIC: Packet Transport Networks

Diunggah oleh

Hak Cipta:

Format Tersedia

Contents

www.zte.com.cn/magazine/English ISSN 1673-5188

Email: magazine@zte.com.cn SPECIAL TOPIC: Packet Transport Networks

Editorial Committee Members (in Alphabetical Order):

SPECIAL TOPIC: Packet Transport Networks

Published and Circulated LECTURE SERIES

Study of the China PTN Industry

private lines/networks. The MANs of 802.1Qay—a connection-oriented

September 2010 Vol.8 No.3 ZTE COMMUNICATIONS 01

following functional requirements for

02 ZTE COMMUNICATIONS September 2010 Vol.8 No.3

September 2010 Vol.8 No.3 ZTE COMMUNICATIONS 03

▼Table 1. OAM requirements in PTN industrial chain is expected to continue

OAM: Operation, Administration and Maintenance VP: Virtual Path

in network deployment, operators must application of PTN technology. The PTN

04 ZTE COMMUNICATIONS September 2010 Vol.8 No.3

September 2010 Vol.8 No.3 ZTE COMMUNICATIONS 05

Engineering Service Instance (TESI). In

06 ZTE COMMUNICATIONS September 2010 Vol.8 No.3

This provides a list of MPLS-TP

September 2010 Vol.8 No.3 ZTE COMMUNICATIONS 07

RFC 5921: A Framework for MPLS in Transport Networks

draft-ietf-mpls-tp-fault: MPLS-TP Fault OAM Refernces

▼Table 2. Individual documents submitted for the 76th IETF meeting

08 ZTE COMMUNICATIONS September 2010 Vol.8 No.3

Several Issues in the Development

September 2010 Vol.8 No.3 ZTE COMMUNICATIONS 09

10 ZTE COMMUNICATIONS September 2010 Vol.8 No.3

September 2010 Vol.8 No.3 ZTE COMMUNICATIONS 11

quickly PTN moves from its introductory

12 ZTE COMMUNICATIONS September 2010 Vol.8 No.3

Features of PBB-TE Architecture

September 2010 Vol.8 No.3 ZTE COMMUNICATIONS 13

PBB-TE discards the spanning tree

14 ZTE COMMUNICATIONS September 2010 Vol.8 No.3

A new label format <B-DA, B-VID>

September 2010 Vol.8 No.3 ZTE COMMUNICATIONS 15

metropolitan area networks: virtual bridged local

16 ZTE COMMUNICATIONS September 2010 Vol.8 No.3

Ring Protection and Survivability

September 2010 Vol.8 No.3 ZTE COMMUNICATIONS 17

18 ZTE COMMUNICATIONS September 2010 Vol.8 No.3

PTN-Oriented Protection and

Some protection mechanisms similar 0.6 0.6

September 2010 Vol.8 No.3 ZTE COMMUNICATIONS 19

B the path message reaches a node, and

Resv 1_i =2 ＜ i 2=4

links needed to establish a path from

20 ZTE COMMUNICATIONS September 2010 Vol.8 No.3

mesh-like capacity for self-planning network

September 2010 Vol.8 No.3 ZTE COMMUNICATIONS 21

Service Adaptation and Label

22 ZTE COMMUNICATIONS September 2010 Vol.8 No.3

September 2010 Vol.8 No.3 ZTE COMMUNICATIONS 23

24 ZTE COMMUNICATIONS September 2010 Vol.8 No.3

(600 in this example).

(VLANs): Ethernet VLAN 1 (port 1A, signaling session

VLAN 100) and Ethernet VLAN 2 (port 1.1.1.1 Transmission 2.2.2.2

September 2010 Vol.8 No.3 ZTE COMMUNICATIONS 25

[6] ITU-T Rec G.8110.1/Y.1370.1. Architecture of

VPN Data Packet VPN Data Packet

CII: Common Interworking Indicators T-LSP: Transport-Label Switched Path

26 ZTE COMMUNICATIONS September 2010 Vol.8 No.3