Anda di halaman 1dari 29

Interim Report

Project Title UBIQUITIOUS SOLUTION FOR MOBILE ADHOC NETWORKS UNDER EMERGENCY CASES

Author Ganesh Ramakrishnan KU Number: K1038457

Course: MSc. Network and Information Security Module: Project Dissertation

Supervisor: Dr. Christos Politis Second Marker: Dr. Maria Martini

TABLE OF CONTENTS 1. Introduction ............................................................................................................................ 3 2. Literature Review.................................................................................................................. 4 2.1 Routing Mechanism .......................................................................................................... 4 2.2 Addressing Mechanism .................................................................................................... 9 2.3 Mobility Management..................................................................................................... 13 2.4 Handoff ............................................................................................................................... 15 3 Methodology .......................................................................................................................... 20 3.1. Scenario Definition ........................................................................................................ 20 3.2. Aims and Objectives ....................................................................................................... 21 4. Initial Results and Discussions ........................................................................................ 21 4.1 The Simulation ................................................................................................................. 21 4.2 Analysis of Simulation Outcome .................................................................................. 21 5. Thesis plan ............................................................................................................................ 23 6. Conclusion and Future Work ........................................................................................... 25 7. References ............................................................................................................................. 26

Abstract
Mobile ad-hoc networks (MANETs) are setup where there is no infrastructure. It is not managed by a central server; it is a joint effort by co-operative nodes to communicate with each other wirelessly. However, under an emergency case, MANETs may be considered suitable to connect and communicate with the world via an internet connectivity provided by the access point. MANETs thus calls for immense consideration to various aspects of routing protocols including proactive, reactive and hybrid routing, IP addressing with consideration to IPv6 mechanisms, mobility management and handovers. A realistic though simple scenario has been considered with discussion to the outcome of the simulation.

1. Introduction
Computers were developed during the peak of industrial era, soon internet was developed and it has grown exponentially since its inception, serving people across different locations of the world. Mobile devices have also advanced to a level where it is capable of computing complex process without much overhead, thus making it easier for such complex devices to connect to the internet. However, there exists such an area which has remained untouched for long time, and that has to be internet connectivity for mobile ad hoc nodes. Mobile Ad-Hoc Networks also known as MANETs have been visualized as flexible, self configurable nodes for communicating amongst one another. MANETs are primarily known for its mobility which may sometimes affect the communication pattern of the nodes. (Perkins et al., 2002). MANETs work in an environment that does not require any infrastructure to be made available. It supports multi-hop communication by using various ad hoc routing mechanisms to transmit information between the nodes (Ratanchandani and Kravets, 2003). MANETS are generally used in those areas which lack any pre-established infrastructure like the Wi-Fi routers or access points, where mobile nodes come within the range of each other to establish an ad hoc connection for a network. Such an infrastructural setup may not be made available due to environmental, natural or technical constraints (Hamidian, 2003, Perkins et al., 2002). But network connectivity can be extended by allowing an Internet Gateway like a router to integrate with the pre-existent mobile ad hoc network to provide internet connection. Thus, enabling total internet service amongst mobile nodes would provide the service exactly in the same way as an ordinarily connected wireless node would have received (Perkins et al., 2002, Ratanchandani and Kravets, 2003). The benefit is that the nodes can not only communicate amongst each other but also to the computers or devices connected to the network remotely via the internet. With mobility being the motif of the document, various design issues will have to be considered in order to unfold a ubiquitous solution for mobility, and seamless movement of mobile nodes.

2. Literature Review
Since the last two decades, there has been an extensive research activity in the field of ad hoc networking. The unpredictable nature of mobile nodes bring can in various challenges that has to be tackled before expecting it to successfully communicate with other nodes. This includes routing protocols, addressing mechanisms, quality of service and seamless mobility (Abolhasan, Wysocki and Dutkiewicz, 2003). Routing in MANETs is a complex task. This is because of the dynamic nature of mobile nodes, where a mobile node primarily routes the traffic, may enter or move away from an ongoing transmission forcing the routing mechanism to condition it for such fluctuations and reconfigure its route to successfully transmit data packets. Thus bringing in delay and affecting the throughput (Abolhasan, Wysocki and Dutkiewicz, 2003, Perkins et al., 2002). This has eventually called upon researchers to design such a routing protocol that could consider changes to the topography of the mobile nodes in any possible scenarios, yet maintain data connectivity and throughput. There are various protocols available that have been designed addressing specific set of scenarios (Abolhasan, Wysocki and Dutkiewicz, 2003). A comparative view of routing protocols has been presented in this section.

2.1 Routing Mechanism


A routing ad hoc routing protocol can be classified as proactive, reactive and hybrid routing protocols (Mobile Ad Hoc Networks (MANET). 2011, Abolhasan, Wysocki and Dutkiewicz, 2003). The Internet Engineering Task Force (IETF) is primarily responsible to standardize the algorithms within European Union (EU). The MANET Working Group of IETF basically standardises new algorithms and promotes further development in the field of MANETs (Mobile Ad Hoc Networks (MANET). 2011). Routing Protocols

Proactive Routing

Reactive Routing

Hybrid Routing

DSDV

WRP

OLSR

AODV

DSR

DYMO

CML

Figure 1: MANET Routing Protocols

2.1.1 Proactive Routing Algorithms


Proactive routing algorithm calculates the route from sender nodes to the destination nodes prior to the transmission of the data packets. However, since the nodes are subject to high level of mobility, there could be a situation where the calculated route may cease to exist while an ongoing transmission (Abolhasan, Wysocki and Dutkiewicz, 2003, Perkins et al., 2002) .

2.1.1.1 Distance Sequenced Distance Vector (DSDV)


Destination Sequenced Distance Vector (DSDV) is a routing table based algorithm with enables a peerless path to the destination. DSDV is inspired from Bellman-Ford shortest path algorithm which based on weighted graph method, creating a shortest path (Abolhasan, Wysocki and Dutkiewicz, 2003, Cormen, Leiserson and Rivest, 2001, Perkins and Bhagwat, 1994). The DSDV algorithm was designed jointly by C. Perkins and P. Bhagwat (Cormen, Leiserson and Rivest, 2001). The DSDV logic is designed in such a way where the routing table assigns a sequence number to the links, switching between even and odd numbers for link state presence or link offline mode respectively. DSDV algorithm sends periodic messages to the network updating the status of the links. This introduces a huge overhead on the network to the effect of O(N2) (Abolhasan, Wysocki and Dutkiewicz, 2003, Cormen, Leiserson and Rivest, 2001). Moreover, this also becomes an overhead for the battery driven mobile nodes. Hence DSDV is not recommended for large ad hoc networks with immense mobility of the nodes (Abolhasan, Wysocki and Dutkiewicz, 2003, Perkins and Bhagwat, 1994). As DSDV is highly inspired by the Bellman-Ford algorithm, DSDV logic also suffers form the problem of count-to-infinity (Abolhasan, Wysocki and Dutkiewicz, 2003). For the selection of the routes, the router considers the most recent sequence number for the link, if the router finds an old entry in the routing table for the same route, then it updates the sequence number, and routes are established (Abolhasan, Wysocki and Dutkiewicz, 2003, Abolhasan, Wysocki and Dutkiewicz, 2003, Perkins and Bhagwat, 1994, Ratanchandani and Kravets, 2003).

2.1.1.2 Wireless Routing Protocol (WRP)


WRP algorithm is inspired from Bellman-Ford algorithm to establish a routing with stale routes information for the same sender-receiver combination (Abolhasan, Wysocki and Dutkiewicz, 2003, Cormen, Leiserson and Rivest, 2001, Murthy and Garcia-LunaAceves, 1996). As compared to DSDV, WRP algorithm tackles the count-to-infinity problem by maintaining authentic route information by using tables and then update network information (Abolhasan, Wysocki and Dutkiewicz, 2003, Murthy and Garcia-Luna-Aceves, 1996). However, creation and maintenance of tables can bring about more memory overhead, and this is directly proportional to the largeness of the network. Similarly, with 5

the increase in the size of the network, WRP algorithm may use a part of available bandwidth to send control packets in order to maintain the connection between the sender and receiver when there is not packet transmission (Abolhasan, Wysocki and Dutkiewicz, 2003, Murthy and Garcia-Luna-Aceves, 1996).

2.1.1.3 Optimised Link State Routing (OLSR)


OLSR is one of the most advanced proactive routing protocols developed by Project Hypercom, INRIA and standardised by IETF (RFC 3626 EXPERIMENTAL). It uses control packets to maintain and manage link information. It is a customised rendition of Link State Routing Protocol suited for MANETs routing. The biggest advantage of OLSR is that it uses Multipoint Relay (MPRs) where only specific nodes relay the network information thus reducing bandwidth usage as compared to the earlier routing protocols discussed in this literature. OLSR introduces Memory Overhead Complexity to the effect of O(N2) and control Overhead complexity of O(N2). This means that OLSR use more memory and network bandwidth. (Abolhasan, Wysocki and Dutkiewicz, 2003, Jacquent and Clausen, 2000, Qayyum, Viennot and Laouiti, 2000).

Figure 2: MPR nodes at 1 hop distance in OLSR (Staub, 2004)

OLSR supports IPv4 and IPv6 and uses User Datagram Protocol (UDP) to transmit control packets and does not rely on any central setup to reliably transmit control information (Abolhasan, Wysocki and Dutkiewicz, 2003, Jacquent and Clausen, 2000).

The main functionalities of OLSR include: 1. 2. 3. 4. 5. 6. Packet Formation and Forwarding Sensing the link by periodic flooding of Hello packets Detecting the neighbour through link sensing Selection of MPR nodes Topology Control (TC) broadcasting Estimating Route Costs

Another benefit of OLSR is that nodes can be pre-configured with the default route information irrespective of the MANET topology, using Host and Network Association (HNA) messages thus enabling total connectivity (Jacquent and Clausen, 2000). However, OLSR does not consider the Quality of Service (QoS) before message transmission, it simply assumes that the link is in its proper state if it is able to transmit Hello packets. It can also be useless such a networking arena where the nodes are offline or asleep after a timeout, for example sensor networks (Abolhasan, Wysocki and Dutkiewicz, 2003, Jacquent and Clausen, 2000).

2.1.2 Reactive Routing Algorithms


Reactive routing protocols algorithms are designed in such a way that would consider the mobility. Therefore, routes are only established when the sender node is ready to transmit. Thus during an ongoing transmission, if a node mobilizes in a manner, where direct transmission of data packets are not possible, routing protocols triggers the route calculation mechanism and establishes a fresh route in no time. Thus, as the name suggests, such routing protocols are reactive in nature (Abolhasan, Wysocki and Dutkiewicz, 2003).

2.1.2.1 Ad Hoc On-Demand Distance Vector (AODV)


AODV is the best available reactive routing protocols jointly designed by the creator of DSDV routing protocol C.E. Perkins of Nokia Research Center, E. Belding-Royer of University of California, Santa Barbara and S. Das from University of Cincinnati in 2003. It has now been standardized by IETF (RFC 3561 EXPERIMENTAL). AODV is a mixture of DSDV and Dynamic Sequenced Routing (DSR) algorithm (Abolhasan, Wysocki and Dutkiewicz, 2003, Perkins, Belding-Royer and Das, 2003). AODV is highly adaptable to movement of the nodes and it is suitable for dynamic MANET topologies. However, with increase in the size of the MANET network, delays may also increase. Hence the route discovery complexity factor for AODV is O(2N) (Abolhasan, Wysocki and Dutkiewicz, 2003). It uses Route Request (RREQ) and Route Reply (RREP) to generate routing information. It may also generate Route Error (RERR) packets to indicate errors in the links. It also uses sequence number for each control packets sent so as to keep a track of fresh route entries in the routing table and differentiate from stale entries. Since AODV is reactive in nature, it stores link information only if the link has been used earlier to the 7

transmission. The route that has been freshly generated is only limited till the end of current transmission. It then may or may not generate a new route depending upon the MANET topology and operating conditions (Perkins, Belding-Royer and Das, 2003). When a node is ready to send a packet, it sends out RREQ packets requesting all the neighbour nodes to respond with identification of their neighbours. Upon the receipt of RREQ packets by the nodes, it has an option of send a response with RREP with the route information pointing towards a particular destination, or it can re-flood the network with RREQ messages (Abolhasan, Wysocki and Dutkiewicz, 2003, Perkins, Belding-Royer and Das, 2003). When there is more than one route towards the destination, AODV will consider the shortest path and the most fresh sequence number indicating the link in order to establish the route (Abolhasan, Wysocki and Dutkiewicz, 2003). RERR messages indicate that the nodes have mobilised from its earlier location, and the sender will have to recalculate the link to the destination (Abolhasan, Wysocki and Dutkiewicz, 2003, Perkins, Belding-Royer and Das, 2003).

2.1.2.2 Dynamic Source Routing (DSR)


DSR algorithm has been standardized by IETF (RFC 4728 EXPERIMENTAL) primarily developed for multi-hop MANET topology. DSR is advantageous for small or medium sized network, where nodes can store more than one link pathway information between sender and receiver. This algorithm is even more useful where mobile nodes rarely mobilize thus routes once established will remain the same until nodes move (Abolhasan, Wysocki and Dutkiewicz, 2003). However DSR implements two operating aspects which are: 1. Route Discovery, where a node intending to transmit a data packet will seek a route to the destination (Abolhasan, Wysocki and Dutkiewicz, 2003, Johnson, Hu and Maltz, 2007). 2. Route Maintenance, indicates that if the node is involved in an ongoing data transmission and it detects a route failure due to the changes in destination due to the change in the network topology, it will trigger the route discovery mechanism to find a fresh route to the destination (Johnson, Hu and Maltz, 2007). DSR algorithm is very robust as it saves multiple routes to a destination from the sender nodes thus making it easier to switch between routes in case if any of the routes breakdown. It uses Address Resolution Protocol (ARP) (Arkko and Pignataro, 2009) for identifying the identity of the next node to the respective MAC Address. ARP can be preconfigured thus reducing the overhead of address resolution broadcast. (Abolhasan, Wysocki and Dutkiewicz, 2003, Johnson, Hu and Maltz, 2007).

2.1.2.3 Dynamic MANET On-Demand (DYMO)


DYMO is a reactive and multi-hop algorithm that operates similar to AODV. It has Route Request (RREQ) and Route Reply (RREP) packets to determine a set of routes (Chakeres and Perkins, 2008). Since DYMO is a multi-hop routing algorithm, when a new route is determined between the sender and a the receiver, and where there is no single hop route available to the destination, the DYMO logic instructs all the intermediate nodes in the route to frame a route towards the destination. This reduces a route discovery operation considerably, since the nodes already have a route to the destination (Chakeres and Perkins, 2008).

2.1.3 Hybrid Routing Algorithms


Hybrid Routing Algorithms are of a different class of routing protocols. They are complex in nature and may sometimes integrate with other existing routing mechanism. Hybrid routing protocols may inherit the advantages of proactive and reactive routing mechanisms, making it dynamic in nature (Abolhasan, Wysocki and Dutkiewicz, 2003).

2.1.3.1Cameleon Routing (CML)


Developed by Wireless Multimedia & Networking (WMN) Research Group of Kingston University London, is an adaptive, scalable routing algorithm that operates considering the MANET topological changes. Since MANET architectures are highly dynamic and fluctuating in nature, CML algorithm analyses the conditions and selects between proactive or reactive routing. It switches between OLSR and AODV for routing thus using the benefits of both the class of routing algorithm to its benefits (Ramrekha, Panaousis and Politis, 2011). CML is designed primarily to address the routing mechanism under emergency situations, where its design requirements include, Battery Limitations, Dynamic mobility of MANETs and maintaining the QoS level in order to achieve better throughput. CML uses a Cognitive and Adaptive Module (CAM) and has three operating phases. CAM module has parameters defined to successfully process o-phase, also known as oscillation phase. In the o-phase the oscillation timings of the node, and then makes a decision of selecting the pphase, proactive-phase and the r-phase, known as reactive-phase (Ramrekha, Panaousis and Politis, 2011). The CAM module decides the selection of phase primarily upon the network size, transmission, delay transmission time and various other parameters. If the MANETs are operating in a small network then it sets OLSR (p-phase), otherwise CAM selects AODV (r-phase) if the network size exceeds a certain threshold (Ramrekha, Panaousis and Politis, 2011).

2.2 Addressing Mechanism


MANET as the name suggest is a ad hoc network of mobile nodes that will require some form of addressing mechanisms in order to identify and communicate with each 9

other. Moreover with the integration of the internet with the MANETs may pose serious challenges to the addressing architecture. The following section will discuss about the challenges and its probable solutions.

2.2.1 Ad-Hoc Network Autoconfiguration (autoconf)


Autoconf working group has been set up to identify and analyse the design requirements of the addressing mechanisms in MANETs. Ad Hoc Networking Working Group of IETF recommends the use of IPv6 addressing schemes for MANETs. Thus the nodes will have to be configured to be able to communicate with IPv6. The Autoconf is responsible to standardise RFC 5889 (Baccelli and Townsley, 5889). Dynamic Host Control Protocol (DHCP) for IPv6 will play a crucial role in Ad Hoc networks. However, any addressing scheme when implemented with DHCP will assume that there is one central node that has a pool of reserve IP addresses which can be assigned to the requesting nodes using DHCP Server. However, such a scenario may not be available in an ever-changing dynamic mobility of MANETs. This is because the node acting as DHCP server may move out of range or go offline while a communication is in progress. Thus there are a lot of opportunities to research upon the unique assignment of IP address to such MANETs (Baccelli, 2008, Baccelli and Townsley, 5889). Autoconf Working Group aims at: 1. Uniquely assign Local and Global IPv4 and IPv6 addresses for MANETs (AdHoc Network Autoconfiguration (autoconf). 2011). 2. Using IPv6 prefixes for MANETs thus making the gateways globally unique (Ad-Hoc Network Autoconfiguration (autoconf). 2011). 3. To establish a method to integrate different networks, but still maintain the same IPv6 prefixes so as to ensure interoperability (Ad-Hoc Network Autoconfiguration (autoconf). 2011). 4. Allow the use of lightweight DHCP, aimed only at assigning non duplicate IPv6 addresses to the MANETs (Ad-Hoc Network Autoconfiguration (autoconf). 2011). There are various design issues with automatic address configurations as discussed in (Baccelli, 2008). They are as follows: 1. Stateless Address Autoconfiguration (SLAAC) may allow automatic assignment of IP address without any server inputs. However, SLAAC depends upon Neighbour Discovery Protocol (NDP) which assumes that each node can communicate with other node in one hop distance which means that each node must be able to communicate with other nodes. However, because of fluctuating nature of MANETs, NDP signalling cannot be used. It will have to be optimised (Baccelli, 2008). 2. Dynamic MANET topology may require unique addresses to be assigned to each node, however with mobility, addressing mechanisms could be 10

disrupted as node move from one router connectivity to another may not belong to the same subnet configuration, resulting in communication fault. This problem may even be amplified if MANETs mobilise between routers at higher rate (Baccelli, 2008). 3. It is recommended that Multiple Gateways could be used, where routers are assigned globally unique prefixes which can allow internet connection to the MANETs (Baccelli, 2008). Autoconf proposes a solution for safe addressing and avoiding IP Address conflict. This solution is known as Full Duplicate Address Detection Model, known as Full-DADModel. There are three phases that has been established under Full-DAD-Model. They are: 1. NO ADDRESS PHASE This is the first stage when a mobile node has been set up. The node has not been assigned any IP Address. The node can self configure its IP Address using any available address generation technique. Once it has an IP Address configured for itself, the node moves to the second stage of the algorithm. However, the node may not be in involved in any data transfer or any other communication (Baccelli, 2008). 2. THE ADVERTISING PHASE In this phase, the mobile node, which recently configured its IP Address, will first check its local database which contains rejected IP Address, if there is an entry of the same address, then algorithm goes back to the first stage of Full-DAD-Model. Alternatively if the algorithm does not encounter the same address in its database, the node advertises its address to the network. Here if it experiences any IP collision, then algorithm alters the database by adding the IP address in conflict and restarts the algorithm (Baccelli, 2008). 3. THE NORMAL PHASE - Once the node is sure that there is no collision of IP address, the algorithm now allows the mobile node to start its normal phase of operation and communicate with neighbouring nodes (Baccelli, 2008).

Full-DAD-Model could be better understood with the following pseudo code.

11

2.2.2 Full-DAD-Model Algorithm


function no_address_phase(void) { sub generate-Address() { //--specific address generation logic call advertising-phase(generated_addr); } } function advertising_phase(generated_addr) { sub pre_DAD_service(generated_addr) { var conflict_detected=0 array collision_detected_ip[]; //-using specific DAD alogithm if(generated_addr in collision_detect_ip[]) { conflict_detected =1; call no_address_phase(void); } else { var addr_exists=Lookup(generated_addr); if(addr_exists=YES) { conflict_detected =1; collision_detected_ip[generated_addr]; call no_address_phase(void); } else { call normal_phase(generated_addr); } } } } function normal_phase(generated_addr) { func start_ip_comms(generated_addr) { // normal communication // in service DAD logic if(duplicate_address detected){ Force no_address_phase(void); } else { Resume normal_communication(); } } }

12

As discussed in address auto-configuration scheme, nodes generate an IP Address from a given IP Address pool. In order to ensure uniqueness, a DAD process is performed. A problem of DAD model as discussed in (Ghosh and Datta, 2011), is that nodes broadcast their address for duplicate address detection, However, if we consider a scenario where numerous nodes try to broadcast their IP Address at the same time, this issue is known as broadcast storm problem (Ni et al., 1999). Moreover, Full-DAD Model may not be suitable for complex network scenarios, where nodes move in and out of the network, thus DADModel may not discover the existence of nodes with an IP Address when the nodes have branched out (Ghosh and Datta, 2011).

2.3 Mobility Management


Mobility is an important factor for performance evaluation of the MANETs. This is because; MANETs mobility may directly influence the aspect of data transmission. Mobility is a pattern of movement, and it has to be analysed before being put to use. Moreover, it has to fit well into the scenario (Staub, 2004). There are plenty of mobility models available for consideration, only a limited number of mobility models are considered in this literature review.

2.3.1 Random Waypoint Mobility


As the name suggests random waypoint mobility refers to the randomised movement of mobile nodes. This model describes the random pattern of the nodes and its location and speed of movement. Random waypoint mobility may generally be used to simulate an open space (Ariyakhajorn, Wannawilai and Sathitwiriyawong, 2006, Bettstetter, Resta and Santi, 2003, Broch et al., 1998, Staub, 2004). Random Waypoint mobility works on timers and speed of the node movement. The mobile node first randomly selects a destination. It then calculates speed of movement based upon a function f(minspeed, maxspeed) as minimum and maximum speeds respectively. Once the mobile node has established all the required parameters it can then start moving towards the destination. At this time, the mobile node pauses and reselects another destination in the space randomly. Thus this process continues till the simulation is complete. By adjusting the velocity of the nodes, performance of the MANETs could change considerably (Bettstetter, Resta and Santi, 2003, Staub, 2004).

13

Figure 3 : Movement Pattern of a node using Random Waypoint Mobility Model (Kraft et al., 2009)

However, there are some disadvantages of using random waypoint model. Such a model is subject to sudden change of paths, unexpected turns and at different velocity irrespective of the prior speeds. Such deviations are generally not foreseen and do not occur in reality. These movement patterns may disturb the average performance evaluation of the nodes and the network. Randomised movement can change the architecture of the MANET networks especially when the velocity of movement is high and the stoppage time is lower as compared to the average pause time (Ariyakhajorn, Wannawilai and Sathitwiriyawong, 2006, Bettstetter, 2001).

2.3.2 Gauss-Markov Mobility Model


The Gauss-Markov mobility model has been put forth by Liang and Haas (Liang and Haas, 1999) and it is one of the popularly available mobility models. This model considers the previous node movement and velocity before calculating the next movement and speed pattern. Additionally, this model casts out the sudden change in speed and direction of the nodes, as seen in random waypoint mobility model. (Ariyakhajorn, Wannawilai and Sathitwiriyawong, 2006, Bettstetter, 2001, Liang and Haas, 1999). This calculation can be understood with the following equation.

Sn Sn1 (1 )S (1 2 ) Sx n1
Equation 1: Node Speed Calculation

d n d n1 (1 )d (1 2 ) d x n1
Equation 2: Node Direction Calculation

14

According to Equation 1 and Equation 2, Sn and dn represents the values of speed and direction respectively. Time is defined as n, and Sn-1 and Dn-1 states the speed and direction of the nodes for the time n-1. This means the speed and direction values of previous unit of time is considered to calculate the next movement pattern. is a numerical constant between 0 and 1 such as 1 0 , representing maximum and minimum node velocity. S and d signifies the average speed and direction of the nodes respectively. S x n 1 and d xn 1 represents output from Gaussian Distribution (Ariyakhajorn, Wannawilai and Sathitwiriyawong, 2006). Figure 4 illustrates the node movement using Gauss-Markov Mobility.

Figure 4: Node Movement pattern using Gauss-Markov Mobility

2.4 Handoff
When there is an aspect of mobility, handovers always comes into the picture. Handoffs or Handovers is an important concept and primary factor to ensure seamless connectivity during mobility. Handovers is a process of handing over the context of an ongoing process. When a node mobilises, the handover algorithm automatically decides to re-associate the mobile node to the nearest available highest signal based upon various parameters (Lim et al., , Montavont and Nol, 2005). During emergency cases, an ad-hoc network will require higher Quality of Service and seamless mobility, thus handover should be as simple and quick as possible. Following are the types of handovers (Dr. Kormentzas and Dr. Sarakis, 2011).

15

1. Horizontal Handover 2. Vertical Handover 3. Inter-domain Handovers 4. Intra-domain Handovers 5. Network initiated handovers 6. Mobile initiated handovers Handovers that involves transferring of context within the same technological background is known as Horizontal handovers; say, handoff between one base station to another of a cellular technology belongs to the category of horizontal handoff. Similarly, handoffs between varied technologies are categorized as vertical handovers. For example, handoff between a 3G cellular network to a Wi-Fi network for communication is deemed to be vertical handover (Dr. Kormentzas and Dr. Sarakis, 2011, Melia et al.) In case of inter-domain handoffs, the transition takes place between different administrative domains. This may sometimes also involve authentication and authorization of the node before associating it to the target domain. Transition from Domain A to Domain B is known as inter-domain handoffs. In case of Intra-domain handoffs, the transition of context takes place within the same administrative domain. For example a cell phone associated to O2 network when roams from point A to point B, it may handoff its context to another cellular base station which is also managed by same network. Thus intra-domain handoffs do not require validation of credentials, and therefore its a faster process (Dr. Kormentzas and Dr. Sarakis, 2011, Melia et al.). Network-initiated handoffs belong to the category of intelligent handoffs. The base station or the network provided is equipped with algorithms to sense the movement of the mobile node, signal power, the location of the mobile node and other service requirements, it triggers handover process. Such a scenario can be found in heterogeneous systems, where a node is allowed to connect to any technology for its communication (Dr. Kormentzas and Dr. Sarakis, 2011, Melia et al., ). In case of mobile initiated handoffs the mobile node is the decision maker for handoff process. It continuously monitors the radio link of all the available interfaces, and based upon various established parameters it then decides to switch between disparate technologies (Melia et al.). When a mobile node moves outside the network area of the current access point, the node will have to connect to the nearest available access point. Such a handover process is known as Layer2 handover. During this phase, the node may not perform any data packets transmission. However, if the target access point is configured in the same way and it also belong to the same subnet group, then the node will can immediately transfer its context to the target access point in a seamless manner. If the access point does not belong to the 16

same subnet, it will have to notify all the routing agents about its new location and its address (Montavont and Nol, 2005, Yokota et al., 2002).

Figure 5 : Ad-Hoc Communication with Router 2

In the above figure the nodes are participating in an ad-hoc communication and node m5 acts as a packet forwarding agent to the main wireless router Router 2. Whereas router Router1 may or may not be involved in some other data transmission. In Figure 6, the mobile node m5 moves and since it the node moves closer to the coverage boundary area, its handover algorithm starts to scan all the nearby channels, and then establish communication with the best available access point. In Figure 7, m5 has finally established connectivity, and the communication continues.

17

Figure 6 : Node mobility and Handover process while Ad-Hoc communication

Figure 7: Ad-Hoc Communication after Handover

18

2.4.1 Handoff Process


Handoff is a three step process, which includes Initialisation, Preparation and Execution. Handoff Initialization includes various parameters that have to be considered including Quality of Service requirements, network resources and Organization policies. The initialisation phase will also have to look into the available targets suitable for handovers (Dr. Kormentzas and Dr. Sarakis, 2011). After this phase, the network and mobile node prepares for handoff, it includes resource reservation in the target network and connection establishment. Figure 6 above proves this phase (Dr. Kormentzas and Dr. Sarakis, 2011). Finally, handover process is executed by updating the location and the address of the mobile node. Figure 7 indicates the scenario after the handover execution (Dr. Kormentzas and Dr. Sarakis, 2011).

Figure 8: Handover Phases (Dr. Kormentzas and Dr. Sarakis, 2011)

19

3 Methodology
For the purpose of simulation, it is recommended to use Network Simulator-2 (NS2) version 2.34 over an Ubuntu 11.04 operating system. An UM-OLSR patch has also been applied on the existing version of NS-2. A TCL file has been scripted in order to simulate the scenario. It contains the definitions for the mobile nodes and various other parameter settings and includes instructions for the simulator as well. Upon successful completion of the simulation, one can analyze the trace files generated and alternatively analyze the simulation in the form of an animation using Network Animator (NAM). A specialized application can be used to create a statistical graph for deeper analysis.

3.1. Scenario Definition


Emergency situations can arise at any time, and therefore a ubiquitous solution can resolve the issues arising out of MANET connectivity with internet connection. Presently, the technology does not permit simultaneous existence of connectivity to ad-hoc network and Wi-Fi connection on the node. There exist several scenarios where MANET internet can prove advantageous 1. Fire outbreak in commercial and residential buildings 2. Natural Disasters including earthquakes, Tornados, Floods, Solar Flares 3. Terrorist Activities in cities 4. Military warfare communication in a city environment 5. Post Disaster Analysis 6. Sensor Networks 7. Students on Campus communication

20

3.2. Aims and Objectives


The aim of the project is to evolve existing technologies in order to enable them to interoperate with each other, thus enabling global communication, keeping in view that MANETs are battery driven.

4. Initial Results and Discussions


4.1 The Simulation
An initial set of simulation was performed with four mobile nodes and two Wi-Fi routers connected to the internet. We assume that these routers are capable enough to operate on dual channel, which means a channel is allocated for Wi-Fi infrastructure and ad-hoc mode. During the simulation, the MANETs initially connect to one of the routers, as we progress, the MANETs mobilise and it connects to the other nearby routers. The simulation runs exactly for 250.00 seconds and uses AODV as default routing protocol. After 2.00 seconds, all the nodes start to transmit data packets. The nodes may also involve in creating routes for data transmission and occasionally associating itself with other routers, should it mobilise. The connection parameters used for simulation is UDP (User Datagram Protocol) with CBR (Constant Bit Rate).

4.2 Analysis of Simulation Outcome


Figure 9 shows the throughput of all four nodes put together. There are two areas in the graph which seems different from the general levels. It is the throughput levels between 40.0 to 60.0 seconds and between 160.0 to 200.0 seconds. In the former case, the nodes mobilise during the aforementioned range of seconds. Node 4 associates with another router and since, there is not much interference, the throughput levels are higher as compared to other nodes which is associated to the older router. During the latter time range, the node mobilise in such a way that the ad-hoc network disintegrates and the nodes connect to the nearest available Wi-Fi routers. The main reason for the drop in the throughput levels is because the nodes are busy in associating itself to the nearest access points. Similarly in Figure 10 below indicates the packets dropped in a given time. Since the node moves and associates with the nearest access points and at times may move out of range, it will lead to drop in packets since nodes are not responsible to ensure data delivery. One of the most noticeable aspects is that Node 4 has the least amount of dropped packets. This is primarily because Node 4 remains at a place free from interference by neighbouring nodes, hence high throughput and low packets dropped.

21

Figure 9: Throughput of 4 Mobile Nodes

Figure 10: Packet Drop for all 4 Nodes

22

5. Thesis plan
Section Introduction Literature Review Content Introduction to MANETs and Generic Overall view of technology Documentation of research work in all related domains, including, addressing, routing, handoffs. Documentation of the simulation methodology including, setting up of the NS-2, patching it with additional support files and scripts. Implementation of proposed solution by simulations and using test-bed Comparison of established generic expected outcome with simulation results or test-bed trace results. Documentation of expected outcome if the project is developed further

Methodology

Implementation Performance evaluation and Discussions

Conclusion and Future Work

23

Figure 11: Gantt Chart including progress

24

6. Conclusion and Future Work


Immense amount of research work has been carried out all over the world in the field of MANETs. With the world becoming more dangerous to live each day, technology can be put to good use if it is allowed to evolve. At this time, researchers from European Union (EU) are pushing ahead for a separate channel allocation in Wi-Fi to support ad-hoc modes under emergency cases all throughout EU. This document can be considered as a stepping stone towards achieving that goal. MANETs are primarily battery driven and operates on limited resources. Routing protocols are considered to be more resource consuming than any other aspect of radio communication. Thus routing protocols have to be tuned to operate under any condition, yet still consume minimal resources and ensure complete throughput. Proactive and reactive algorithms have their share of disadvantages, and the advantages of both have been put together to develop hybrid routing protocols. Addressing plays a crucial role in ensuring total connectivity. Various models have been proposed and none of them provides a fool-proof solution for IP addressing. Moreover, IPv4 addresses are fast depleting and IPv6 can resolve addressing issues. Nodes can move in any way they want challenging connectivity amongst other nodes or with the main access point, which calls upon handoff mechanism which puts light upon seamless mobility. Seamless mobility requires handing over the context of the connection from one access point to another. The search for a ubiquitous solution is still ongoing, and researchers are expected to resolve the addressing issues, and solve the subnetting mechanism especially where nodes move from one subnet to another, triggering a change in the address space.

25

7. References
Ad-Hoc Network Autoconfiguration (autoconf). (2011) Available th http://datatracker.ietf.org/wg/autoconf/charter/ IETF. (Accessed on: 14 /October/2011). at:

Mobile Ad Hoc Networks (MANET). (2011) Available at: http://datatracker.ietf.org/wg/manet/charter/ IETF. th (Accessed on: 11 /October/2011). Abolhasan, M., Wysocki, T. and Dutkiewicz, E. (2003) 'A review of routing protocols for mobile ad hoc networks', Ad Hoc Networks, . Ariyakhajorn, J., Wannawilai, P. and Sathitwiriyawong, C. (2006) A Comparative Study of Random Waypoint and Gauss-Markov Mobility Models in the Performance Evaluation of MANET. Arkko, J. and Pignataro, C. (2009) IANA Allocation Guidelines for the Address Resolution Protocol (ARP). th Available at: http://tools.ietf.org/html/rfc5494 IETF. (Accessed on: 14 /October/2011). Baccelli, E. (2008) Address Autoconfiguratuion for MANET: Terminology and Problem Statement. MANET Autoconfiguration (Autoconf) - Internet Draft. IETF. Available At: http://wiki.tools.ietf.org/html/draft-ietfth autoconf-statement-04(Accessed on: 14 /October/2011). Baccelli, E. and Townsley, M. (5889) IP Addressing model in Ad Hoc Networks - RFC 5889. IETF AD Hoc th Working Group. IETF. Available At: http://tools.ietf.org/html/rfc5889(Accessed on: 17 /October/2011). Bettstetter, C. (2001) Smooth is better than sharp: a random mobility model for simulation of wireless networks. ACM. Bettstetter, C., Resta, G. and Santi, P. (2003) 'The node distribution of the random waypoint mobility model for wireless ad hoc networks', Mobile Computing, IEEE Transactions on, 2 (3), pp. 257-269. Broch, J., Maltz, D., Johnson, D., Hu, Y. and Jetcheva, J. (1998) 'A Performance comparison of Multi-hop wireless Ad Hoc Network Routing Protocols', pp. 85-97. Chakeres, I. and Perkins, E., C. (2008) Dynamic MANET on-Demand (DYMO) Routing. MANET Working Group Internet - Draft. IETF. Available At: http://ianchak.com/dymo/draft-ietf-manet-dymo-12.txt. th (Accessed on 16 /October/2011) Cormen, T., H., Leiserson, C., E. and Rivest, R. L.(2001) 'Section 24.1: The BellmanFord algorithm', in 'Section 24.1: The BellmanFord algorithm', in Introdution to Algorithms - Second Edition. MIT Press and McGraw-Hill, pp. 588-592, 614-615. Dr. Kormentzas, G. and Dr. Sarakis, L. (2011) Vertical Handovers in Heterogeneous Next-Generation Wireless Networks. Kingston University. Ghosh, U. and Datta, R. (2011) 'A secure dynamic IP configuration scheme for mobile ad hoc networks', Ad Hoc Networks, 9 (7), pp. 1327-1342. Hamidian, A. (2003) 'A simulation of internet connectivity for mobile ad hoc networks in ns2',

26

Jacquent, P. and Clausen, T. (2000) Optimised Link State Routing Protocol - RFC 3626. Request for Comments: 3626. 03rd edn. IETF Network Working Group. Available At: th http://tools.ietf.org/html/rfc3626(Accessed on: 13 /October/2011). Johnson, D., Hu, Y. and Maltz, D. (2007) The Dynamic Source Routing Protocol (DSR) for Mobile Ad hoc Networks for IPv4 - RFC 4728. IETF. Available At: http://www.ietf.org/rfc/rfc4728.txt(Accessed on: th 13 /October/2011). Kraft, D.Bechler, M.Hof, H.Pahlke, F. and Wolf, L. (2009) 'Design and evaluation of a security architecture for ad hoc networks' International Journal of Pervasive Computing and Communications, Available at: th http://dx.doi.org/10.1108/17427370911008848 (Accessed on: 20 /October/2011). Liang, B. and Haas, Z. J. (1999) Predictive distance-based mobility management for PCS networks. IEEE. Lim, W. S., Kim, D. W., Kim, W. J., Suh, Y. J., Cha, Y. M. and Chung, B. D. Design of Multi-Hop Handovers for Hybrid Ad Hoc Networks in Multi-Channel Environments. IEEE. Melia, T., Korhonen, J., Aguiar, R., Sreemanthula, S. and Gupta, V. Network Initiated Handovers. Available At: www.ietf.org/proceedings/66/slides/mipshop-5/mipshop-5.ppt (Accessed on: st 21 /October/2011). Montavont, N. and Nol, T. (2005) Anticipated handover over IEEE 802.11 networks. IEEE. Murthy, S. and Garcia-Luna-Aceves, J. J. (1996) 'An efficient routing protocol for wireless networks', Mobile Networks and Applications - Special Issue: Routing in Mobile Communications Networks, 1 (2), pp. 183-197. Ni, S. Y., Tseng, Y. C., Chen, Y. S. and Sheu, J. P. (1999) The broadcast storm problem in a mobile ad hoc network. ACM. Perkins, E. C., Malinen, T. J., Wakikawa, R., Nilsson, A. and Tuominen, J. A. (2002) 'Internet Connectivity for Mobile Ad Hoc Networks', Wireless Communication and Mobile Computing, pp. 465-482. Perkins, E., C., Belding-Royer, E. and Das, S. (2003) Ad hoc On-Demand Distance Vector (AODV) Routing RFC 3561. IETF-NETWORK WORKING GROUP. IETF. Available At: th http://www.ietf.org/rfc/rfc3561.txt (Accessed on: 13 /October/2011). Perkins, E., C. and Bhagwat, P. (1994) 'Highly Dynamic Destination-Sequenced Distance-Vector Routing (DSDV) for Mobile Computing', ACM SIGCOMM, pp. 234-244. Qayyum, A., Viennot, L. and Laouiti, A. (2000) 'Multipoint Relaying: An efficient technique for flooding in mobile wireless networks.', INRIA ROCQUENCOURT, . Ramrekha, A. T., Panaousis, E. and Politis, C. (2011) Cameleon (CML): A hybrid and adaptive routing protocol for Emergency Situations. IETF MANET Working Group- Internet Draft (Experimental). 02nd edn. IETF. Available At: http://tools.ietf.org/html/draft-ramrekha-manet-cml-02 (Accessed on: th 14 /October/2011). Ratanchandani, P. and Kravets, R. (2003) 'A Hybrid Approach to Internet Connectivity for Mobile Ad Hoc Networks', Wireless Communications and Networking, IEEE, .

27

Staub, T. (2004) Performance Analysis of MANET Routing Protocols in Ad-Hoc and Hybrid Networks. Masters. University of Berne, Switzerland. Yokota, H., Idoeu, A., Hasegawa, T. and Kato, T. (2002) 'Link Layer Assisted Mobile IP Fast Handoff Method over Wireless LAN Networks', pp. 1-9.

28

Anda mungkin juga menyukai