I NTRODUCTION
ITH almost 6 billion mobile subscribers worldwide [1], mobile phones have become ubiquitous
devices. The increasing popularity of iOS- and Androidbased applications has made mobile phones with richer features an indelible part of modern popular culture. However,
the limited battery life of mobile phones constrains the
emergence of more attractive and complex applications.
In cellular networks, radio resources shared among user
equipments (UEs, e.g., smartphones) are known to be the
major power consumer in mobile phones [16], [32], [34].
However, a large proportion (nearly 60%) of energy consumed by radio resources is derived from the timeout
period of inactivity timers [23]. These timers are used to
control the release of radio resources [7]. The idle interval corresponding to the inactivity timer timeout period
before radio resources are released is referred to as the
tail time, and it is adopted to balance the trade-offs among
radio resources, user experience, energy consumption, and
network overhead [26]. However, the tail time results in
the wastage of radio resources and battery energy in
UEs [23], [25].
We focus on the Universal Mobile Telecommunications
System (UMTS) network, one of the most popular 3G
mobile communication technologies. To use limited radio
resources efficiently [12], the UMTS introduces the Radio
c 2013 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
1536-1233
See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
ZHANG ET AL.: LEVERAGING THE TAIL TIME FOR SAVING ENERGY IN CELLULAR NETWORKS
1537
Fig. 1. (a) Overview of the RRC state machine. (b) Power curve of an
example transfer over the UMTS network.
BACKGROUND
The Radio Network Controller (RNC) is an important component of the UMTS network. Most features of the UMTS
Terrestrial Radio Access Network such as radio resource
management, packet scheduling, and some mobility management functions are implemented in the RNC. Radio
resources shared among UEs are potential bottlenecks in
the network. To use the limited resource efficiently, the
RRC protocol described in [7] maintains a state machine
(Fig. 1(a)) for both the UE and RNC.
Typical RRC states are described as follows.
CELL_DCH. In the CELL_DCH (Dedicated Channel,
DCH hereafter) state, a dedicated physical channel is allocated to the UE in both downlink and uplink, thus enabling
the UE to use radio resources fully for the high-speed
transmission of user data. Given that communication via
DCH is bidirectional, both the transmitter and receiver of
the UE should be active, thereby resulting in high power
consumption in this state.
CELL_FACH. In the CELL_FACH (Forward Access
Channel, FACH hereafter) state, no dedicated physical
channel is allocated to the UE. Instead, the UE uses the
shared channels FACH (downlink) and Random Access
Channel (uplink) to transmit control messages and a small
amount of user data at very low speeds. Given that
FACH uses shared channels, this state consumes less radio
resources and battery energy.
IDLE1 . In this state, no RRC connection is established,
and no radio resource is allocated. The UE cannot transfer
user data.
Promotion and demotion are the two types of state
transitions. As illustrated in Fig. 1(a) and (b), the UE consumes more (less) radio resources and power after promotion (demotion). Promotion, which includes IDLEDCH,
1. A hibernation state called CELL_PCH is found in some UMTS
networks. It is similar to IDLE, but the state promotion delay is shorter.
1538
P RELIMINARY F ORMULATION
ZHANG ET AL.: LEVERAGING THE TAIL TIME FOR SAVING ENERGY IN CELLULAR NETWORKS
its arrival ai and before its deadline di , but should be discarded as its deadline approaches. When i is transmitted at
time si , the RRC state machine generally transits to a highpower state (it may also transit to an intermediate-power
state) and transmits i instantaneously. After transmission
is completed and no additional transmission occurs, the
state machine remains in the high-power state for T1 time
units before transiting to a intermediate-power state. If
no transmission occurs, the state machine remains in the
intermediate-power state for T2 time units before transiting to the lower-power state, where T1 and T2 are
the tail time. Even when multiple requests are simultaneously transmitted, state transition remains the same
with only one request. Let E(S) denote the total energy
consumption of the schedule S. The request scheduling
problem is to compute a feasible request schedule S that
minimizes E(S).
A simplistic schedule treats all requests as real-time,
without considering delay and prefetch characteristics.
However, this schedule may result in more state transitions and longer tail time. If numerous requests need to be
scheduled, substantial energy is consumed by the tail time
and state promotions, and is also drained by sparse transmission. Additionally, given that the request queue is not
known a priori in practice, the schedule has to be computed
online. TailTheft follows the principle that delay-tolerant
requests and previous attempts are delayed and transmitted in the tail time within their deadlines. This action not
only uses the tail time, but also reduces transmission time
and numerous promotions.
1539
1540
t2D
t1D
Dw (vt , t)dt +
t2F
t1F
ZHANG ET AL.: LEVERAGING THE TAIL TIME FOR SAVING ENERGY IN CELLULAR NETWORKS
maintained for data requests: one for real-time and unsuccessfully prefetched prefetchable requests and another for
delay-tolerant requests and previous attempts. We refer to
delay-tolerant requests and previous attempts as TailTheft
requests. When a request is added to the TailTheft request
queue, TailTheft starts a timer , the timeout value of which
is the latest deadline of all the requests in the queue. Timer
ensures that all delayed requests are processed before the
specified deadline.
1541
TABLE 2
Parameters Set in TailTheft Implementation
1542
S IMULATIVE I MPLEMENTATION
ZHANG ET AL.: LEVERAGING THE TAIL TIME FOR SAVING ENERGY IN CELLULAR NETWORKS
TABLE 3
E-mail Trace
1543
E VALUATION
We evaluate TailTheft performance against that of two existing representative schemes, TailEnder [23] and TOP [25],
where TailEnder is a traffic aggregation scheme and TOP
is a tail time tuning scheme. TailEnder and TOP are also
implemented in NS-2, but for simplicity, TOP is implemented with a prediction accuracy of upcoming non-tail
equal to 100% and a prediction accuracy of upcoming tail
equal to p. These prediction accuracies result in better performance of TOP in this paper compared with that of
previous study [25].
Two kinds of requests, delay-tolerant and prefetchable,
are used to evaluate TailTheft, because a number of requests
can be deferred or prefetched, and these requests are the
key targets of TailTheft. According to the analysis of Falaki
et al. [31], [32], we select requests of two common applications, e-mail and news, for evaluation. E-mail is an
application that can tolerate a moderate delay. News is an
application that can benefit from prefetching. In addition,
delay-tolerant and prefetchable requests are always mixed
with real-time ones. Therefore, the performance with mixed
requests is also evaluated.
We use E(S), D(S), F(S), P(S) and A(S) (defined
in Section 3.3) as evaluation metrics. These metrics are calculated using the energy consumption models proposed in
Section 3.4 and the parameters established in Section 5. The
comparison baseline is the default transmission schedule
with unoptimized tail time.
1544
TABLE 4
News Trace
Fig. 5. Impact on resource metrics for delay-tolerant requests with varying delay deadlines. (a) E(S). (b) D(S). (c) F (S). (d) P(S).
(e) A(S). (f) E(S) of different users.
ZHANG ET AL.: LEVERAGING THE TAIL TIME FOR SAVING ENERGY IN CELLULAR NETWORKS
1545
Fig. 6. Impact on resource metrics for prefetchable requests with varying prefetch accuracies. (a) E(S). (b) D(S). (c) F (S). (d) P(S).
(e) A(S). (f) E(S) of different topics.
1546
Fig. 7. Impact on resource metrics for mixed requests with different ratios of real-time requests. (a) rar = 0.5. (b) rar = 0.2. (c) rar = 0.8.
R ELATED W ORK
ZHANG ET AL.: LEVERAGING THE TAIL TIME FOR SAVING ENERGY IN CELLULAR NETWORKS
feature, TailEnder [23] performs batching and prefetching for e-mail and web search, respectively, whereas
Cool-Tether [17] performs aggregation for browsing. The
low accuracy of the prediction of future transmissions
may cause unnecessary energy consumption, which is the
greatest disadvantage of these two approaches. Although
TailTheft is also based on aggregatable applications, it conducts batching and prefetching in the tail time. Even in
a worst-case scenario, TailTheft does not increase energy
consumption.
Tail Time Tuning. A number of previous studies on
optimizing inactivity timers are based on tail time tuning.
It is suggested that a tail time of 10 seconds to 15 seconds is a reasonable range for controlling the probability
of state promotion to below 5% [18]. Researchers [21], [20]
propose analytical models for investigating the effect of different timer values on energy consumption in UEs. It is
shown that setting the tail time to 4.5 seconds significantly
reduces energy consumption [31]. Although these studies
identify a reasonable tail duration for saving energy, the
tail time still exists and causes substantial energy wastage.
TOP [25], RadioJockey [29] and Traffic-aware approach [30]
are three recently proposed methods to tune the tail time.
TOP and RadioJockey utilize a feature called fast dormancy [11] to terminate the tail time dynamically if they
predict that no data requires further transmission. The
difference between TOP and RadioJockey lies in their prediction methods: TOP provides applications with an API
to actively invoke fast dormancy, whereas RadioJockey uses
program execution traces to predict the end of communication spurts to similarly invoke fast dormancy. Traffic-aware
approach dynamically brings the radio to a connected or
idle state by learning traffic patterns and predicting the
start and end of traffic spurts, and it also invokes fast dormancy to bring the radio to an idle state. However, the
efficiency of dynamic tuning of these methods depends
on the accuracy of the prediction of future traffic patterns. Low prediction accuracy not only retains untuned
tail time that still wastes energy, but also causes additional
state promotions, thereby wasting more energy. In addition,
both Traffic-aware approach and RadioJockey are designed
only for background traffic and thus cannot reduce energy
wasted in the tail time generated by user interactive applications. Furthermore, tail time tuning methods consume
lots of energy in transmitting lengthy and sparse traffic
because the radio resource state machine is forced to remain
in a high-power state for a long time. By batching and
prefetching in the tail time, TailTheft utilizes the unused
tail time of both background and interactive applications.
Moreover, batching and prefetching in the tail time can
reduce the total transmission time, thereby saving more
energy.
In addition to mitigating the tail time effect, saving
radio energy in UEs has also been considered. Given that
cellular radios consume more energy when signals are
weak, researchers [37] develop an energy-aware cellular
data scheduling system. In [38], the authors determine
energy-delay trade-offs in smartphone applications and [34]
discuss user activity patterns for saving energy. Other
mobile energy saving technologies, such as WiFi [39], [40]
and GPS [41], have also been studied.
1547
1548
ACKNOWLEDGMENTS
The authors would like to thank the anonymous reviewers
for their constructive comments which helped improve the
quality of the manuscript significantly. The work is partially
supported by the National Natural Science Foundation of
China (Grant 60903029).
R EFERENCES
[1] The World in 2011: ICT Facts and Figures [Online]. Available:
http://www.itu.int/ITU-D/ict/facts/2011/index.html
[2] Nokia Energy Profiler [Online]. Available: http://store.ovi.com.cn/
content/73969
[3] The Network Simulator [Online]. Available: http://www.isi.edu/
nsnam/ns
[4] Enhanced UMTS Radio Access Network Extensions for NS-2 [Online].
Available: http://eurane.ti-wmc.nl/eurane
[5] TailTheft [Online]. Available: https://github.com/dizhang/
tailtheft
[6] GERAN RRC State-Machine, 3GPP TSG GERAN Adhoc #2 GAHW(00)0027, 2000.
[7] Radio Resource Control (RRC); Protocol Specification (Release 8), 3GPP
TS 25.331 V8.0.0, 2007.
[8] System Impact of Poor Proprietary Fast Dormancy, TSG-RAN meeting #45 RP-090941, 2009.
[9] Technical Specifications and Technical Reports for a UTRAN-based
3GPP system (Release 7), 3GPP TS 21.101 V7.0.0, 2007.
[10] Technical Specifications and Technical Reports for a UTRAN-Based
3GPP System (Release 8), 3GPP TS 21.101 V8.0.0, 2009.
[11] Interlayer Procedures in Connected Mode (Release 8), 3GPP TS 25.303
V8.0.0, 2007.
[12] H. Holma and A. Toskala, HSDPA/HSUPA for UMTS: High
Speed Radio Access for Mobile Communications, Chichester, U.K.:
Wiley, 2000.
[13] S. Sesia, I. Toufik, and M. Baker, LTE, The UMTS Long Term
Evolution: From Theory to Practice, Chichester, U.K.: Wiley, 2009.
[14] M. Chatterjee and S. K. Das, Optimal MAC state switching for
CDMA2000 networks, in Proc. IEEE INFOCOM, New York, NY,
USA, Nov. 2002, pp. 400406.
[15] H. Liu, Y. Zhang, and Y. Zhou, TailTheft: Leverageing the wasted
time for saving energy in cellular communications, in Proc. ACM
MobiArch, Bethesda, MD, USA, Jun. 2011, pp. 3136.
[16] A. Carroll and G. Heiser, An analysis of power consumption in
a smartphone, in Proc. USENIX ATC, Berkeley, CA, USA, Jun.
2010, pp. 114.
[17] A. Sharma, V. Navda, R. Ramjee, V. N. Padmanabhan, and
E. M. Belding, Cool-tether: Energy efficient on-the-fly WiFi hotspots using mobile phones, in Proc. ACM CoNEXT, Rome, Italy,
Dec. 2009, pp. 109120.
ZHANG ET AL.: LEVERAGING THE TAIL TIME FOR SAVING ENERGY IN CELLULAR NETWORKS
1549
Yuezhi Zhou received the Ph.D. degree in computer science and technology from Tsinghua
University, China, in 2004 and is now an
Associate Professor in the same Department. He
was a visiting scientist at the Computer Science
Department in Carnegie Mellon University in
2005. His current research interests include
ubiquitous/pervasive computing, distributed system, mobile device and systems. He has published over 60 technical papers in international
journals or conferences. He received the IEEE
Best Paper Award in the 21st IEEE AINA International Conference in
2007. He is a member of the IEEE and ACM.