IJISET - International Journal of Innovative Science, Engineering & Technology, Vol. 2 Issue 2, February 2015.

www.ijiset.com
ISSN 2348 – 7968

Dynamic Packet Length Optimization in Wireless Sensor Network

1
Mr. Atul M. Lade, 2Mr. Sanjay S. Dorle

M.E. Student, Mobile Technology GHRCE, Nagpur, India

Associate Professor, Department of Electronics Engineering, GHRCE, Nagpur, India

Abstract— In wireless sensor networks, dynamic packet can lead to staggered collisions, which occur when node A’s
length control is more efficient in terms of channel packet is interrupted by the hidden node, causing reception to
utilization and energy efficiency. In previous packet length fail. The risk of this type of loss increases with packet length
optimizations for sensor network often employ a fixed because as length increases, the hidden node is required to
optimal packet length scheme, while dynamic packet remain silent for a longer period of time. Additionally, longer
length provide accurate radio link estimation and increase packets are more susceptible to loss due to channel errors
system throughput. The adaptation of dynamic packet because they require the successful decoding of more
length is 802.11 wireless system. The packet delivery ratio symbols[10].
keeps high i.e. 95% above and link estimated error within Packet length adaptation seeks to address the tradeoff between
10% for 95% link. The experiments provide optimization lower overhead for long packet lengths, and lower probability
of dynamic packet length achieves best performances of loss for shorter packet lengths. There has been a significant
related to the previous experiments. amount of research on packet length adaptation. In current
packet length adaptation literature, a simple packet loss model
Index Terms—Link estimation, packet length is typically used, assuming the channel to have a constant bit-
optimization, wireless sensor networks. error rate (BER), and neglecting staggered collisions [1, 2].
I.INTRODUCTION This assumes that most packet losses occur due to random bit
errors in the packet payload. However, it has experimentally
A fundamental challenge in wireless networks is that radio been shown that for lower modulation rates in 802.11a, most
links are subject to transmission power, fading, and packet losses occur due to failure to synchronize to the packet
interference, which degrade the data delivery performance. preamble . This type of loss cannot be accounted for using a
This challenge is exacerbated in wireless sensor networks constant BER model, as it requires a model of channel
(WSNs), where severe energy and resource constraints fading[2].
preclude the use of many sophisticated techniques that may be The purpose of this study is to find the optimal packet length
found in other wireless systems [1][3].For example, (i) bit rate for the real time channel conditions. The basic idea is: if the
adaptation protocols, demand special hardware that is not packet length is too large, the packet retransmission rate will
available on general sensor nodes; (ii) effective forward error be high due to packet error; on the other hand, if the packet
correction (FEC) requires the amount of redundant data length is too small, the effective data throughput will be low
transmitted to be tuned to match the link qualities which is since much transmission is spent on packet headers. Therefore
difficult to achieve in dynamic WSNs[3]. some optimal packet length exists to achieve maximal
802.11 wireless LANs were originally designed for small throughput[10].
networks with limited traffic, and are thus not optimized for A packet optimization scheme applicable in sensor networks
high traffic situations. However, as wireless LANs become must have the following important features.
increasingly ubiquitous, the design limitations become greatly (i) Dynamic packet length adaptation. Prior work in sensor
stressed. One often neglected tunable parameter is MAC layer networks uses fixed packet length optimization schemes.
packet length. While packet length can be variable in the These schemes are not preferred due to the spatial temporal
802.11 standard, it is most often simply set to the maximum diversity of link qualities in WSNs.
value to reduce the impact of overhead. This is indeed the (ii) Accurate link estimation. The performance improvement
optimal setting for a scenario with a single pair of nodes with of the packet length adaptation scheme is highly dependent on
a strong channel; however, in scenarios with hidden nodes the link estimation accuracy. Prior work in wireless systems
and weaker channels, shorter packets may be preferable due does not consider unique characteristics of WSNs, e.g.,
to their lower susceptibility to loss. A hidden node for a given resource constraints of sensor nodes, thus leading to
node A transmitting to a node B in a wireless network is one inaccurate link estimation in sensor networks.
which is capable of interfering with the reception of the packet (iii) Easy to use. To the best of our knowledge, no prior work
at B, while being unable to sense the transmission of A. This

193
 IJISET - International Journal of Innovative Science, Engineering & Technology, Vol. 2 Issue 2, February 2015.
www.ijiset.com
ISSN 2348 – 7968
addresses the application programmability issues of packet Figure. 1. Throughput vs packet length assuming no collisions
length adaptation scheme in sensor networks. Without for (a) SNR fixed at 9dB, and (b) SNR with mean of 9dB with
substantial programming efforts, there is still a huge gap standard deviation of 3dB.
between theoretical optimality and practically achievable
gains. B) Dynamic Packet Length Adaptation:
The rest of the paper is organized as follows. Section II
discussed related work. In section III comparative analysis of Wireless link qualities can be greatly affected by
various techniques with the help of table. In section environmental factors at different locations . To see how link
IV research design is explained. Finally dynamic packet diversity is, we setup one TelosB mote to measure the
optimization scheme it proposes. received signal strength at a resolution of 1 ms. The
experiments are conducted both indoor (the environment is
noisy because of 802.11 interferences) and outdoor (the
II.RELATED WORK environment is quiet), respectively. We can see that the
channel conditions vary drastically: in the noisy environment
A. A) Packet Loss Model:
the RSSI value can be as high as −62 dbm while in the quiet
environment the RSSI value can be as low as −96 dbm.
Losses in Wireless LANs can be broadly classified into two This indicates that packet length adaption schemes must be
types: collisions, which are the result of unfavorable traffic dynamically adapted to physical channel conditions to deliver
conditions, and channel errors, which are the result of performance gains in WSNs with spatial-temporal diversity in
unfavorable channel conditions. A collision occurs when a link qualities.
node’s packet overlaps in time with that of another node
which is spatially close enough to the destination to interfere. C) Accuracy of Link Estimation:
A channel error occurs when the SNR of a received packet is
low due to a large path loss or a deep multipath fade. The total We evaluate the accuracy of the link estimation method
packet loss probability PL can be computed as: in terms of absolute error. Each node transmits 200 packets
in turn. All other nodes record each packet’s reception. The
PL = 1− (1 − PC)(1 − Pe) estimated link qualities are compared against the ground
truth values. Figure 8 shows the CDF of absolute error (i.e.,
where PC is the probability of collision, and Pe is the estimated value - real value). We can see that the errors keep
probability of channel error, which is assumed to be within 10% for 95% links, indicating that the link estimation
independent of PC. method yields accurate results.
In this analysis, we assume that all collided packets are lost,
not captured, and that the probability of ACK loss is
negligible compared to other losses.
Figure 1. Figure 1(a) assumes a constant BER, and
Figure 1(b) assumes SNR to have a log-normal distribution.
The shapes of the curves are noticeably different. This is
because in the case where SNR is modeled probabilistically,
the actual value of SNR has a much higher impact on an
individual packet’s successful transmission than the packet’s
length. While a constant BER model might suggest using a
packet length of only 400 bytes, a more accurate model
including SNR distribution shows that maximum packet
length would be superior.

Figure 2. Accuracy of link estimation.

194
 IJISET - International Journal of Innovative Science, Engineering & Technology, Vol. 2 Issue 2, February 2015.
www.ijiset.com
ISSN 2348 – 7968

III.COMPARATIVE ANALYSIS

Paper Packet Loss Average Delay(ms) Bit- Distortion−rate Scheme /Algorithms

performance for
streaming(b/s)

Packet Data 7.01 % 314 124 Network path

Transmission Over diversity, streaming of
Mobile packetized length.
Radio Channels

Dynamic Packet 9.13% 421 122 Error-cost optimized

Length Control in transmission of a
Wireless single data
Sensor Networks unit in isolation.

Packet Length 5.21% 214 124 Multiple access

Adaptation in WLANs networks, integer
with Hidden programming
Nodes and Time- problem.
Varying Channels

Adaptive Design for 2.12% 131 54 Cross-layer

the Packet Length of optimization, optimal
IEEE packet length,
802.11n Networks distributed
coordination
function (DCF).

IV. RESEARCH DESIGN accurate link estimation method that can capture both
physical channel conditions (due to channel fading, mobility,
In this section, we present the design of DPLC, a dynamic or power degradation) and interferences (from
packet length control scheme for WSNs. Below, we identify exposed and hidden terminals).
the major design goals. • Ease of programming. DPLC should provide easy-touse
• Dynamic adaptation. DPLC should provide a dynamic services to facilitate upper-layer application programming.
adaptation scheme to achieve performance improvements • Lightweight for implementation. DPLC should be
in dynamic, time-varying sensor networks. lightweight for resource constrained sensor nodes.
• Accurate link estimation. DPLC should implement an

195
 IJISET - International Journal of Innovative Science, Engineering & Technology, Vol. 2 Issue 2, February 2015.
www.ijiset.com
ISSN 2348 – 7968
A. Flow chart: Both AS∞ and ASn requires L2 ACKs provided by the link
layer, because packets need to be retransmitted (at least once)
The flowchart of dynamic packet length optimization when they are lost. For AS0, we additionally provide a more
approach is described with the following diagram: efficient ACK scheme called AggAck that does not rely on L2
ACKs, and thus mitigate the ACK overhead (we use AS0-L2
to denote AS0 with L2 ACKs and AS0-AA to denote AS0 with
AggAck afterwards).
(ii) FS is useful for bulk data transmission, e.g., Flush.
FS provides reliable transmissions as a large message is
usually very important for upper-layer applications. FS does
not necessarily depend on L2 ACKs. As mentioned above, we
additionally provide the AggAck mechanism to mitigate the
ACK overhead, and more importantly, to deal with data
packet retransmissions (we use FS-L2 to denote FS with L2
ACKs and FS-AA to denote FS with AggAck afterwards).

V. IMPACT OF DYNAMIC OPTIMIZATION MECHANISM

This mechanism proposed here will provide the
accurate link estimation, efficient utilization of channel in
outdoor environment. This will improve the system
throughput and packet delivery ratio in wireless sensor
Figure 3: DPLC Overview networks.
VI. EXPECTED OUTCOME
B. Proposed Methodology This dynamic packet length optimization appraoach
provide the packet delivery ratio keeps high i.e. 95% above
and link estimated error within 10% for 95% link.
The DPLC scheme works as follows. The application passes
an application-level message for transmission. The DPLC
module at the sender decides whether to use the aggregation VII. ACKNOWLEDGEMENT
service (AS, if the message length is small) or the
fragmentation service (FS, if the message length is larger than I am currently pursuing M.E in Mobile Technology and the
the maximum packet length supported by the radio, i.e., 128 work in this paper is under process. The dynamic packet length
bytes for CC2420). The link estimator within DPLC optimization approach will be implemented using the software
dynamically estimates the appropriate packet length for modules. Further, dynamic packet length optimization
transmission. Based on this, the DPLC module at the sender approach will be used for accuracy of link estimation, efficient
decides how many messages should be aggregated (for AS), or channel utilization and increase system throughput.
how many frames the message should be fragmented into (for VIII. CONCLUSION
FS). When a frame is ready for transmission (enough
messages have been aggregated or time is out in AS), DPLC This proposed dynamic packet length optimization
transmits it out via the MAC layer. When the DPLC module at approach will provide accuracy in link estimation that capture
the receiver receives a MAC frame, it deaggregates or physical channel condition, increase packet delivery ratio,
defragments the frame in order to obtain the original message. increase system throughput and efficient energy utilization.
When the message is ready (all frames in the message have
been received or the receive buffer is full in FS), the DPLC REFERENCES
module at the receiver notifies the upper layer for further
handling. The DPLC scheme provides two services for upper- [1]Wei Dong, Chun Chen, Xue Liu, Yuan He,Yunhao Liu,
layer applications, i.e., the aggregation service (AS, for small Jiajun Bu, Xianghua Xu, “ Dynamic Packet Length Control in
messages) and the fragmentation service (FS, for large Wireless Sensor Network,” in Proc. 2014 IEEE Transactions.
messages). [2] M. N. Krishnan, E. Haghani, and A. Zakhor, “Packet
(i) AS is useful for small data collection, e.g., CTP [7]. length adaptation in WLANs with hidden nodes and time-
AS provides three distinct mechanisms, i.e., reliable varying channels,” in Proc. 2011 IEEE GlobeCom.
transmissions (AS∞), unreliable transmissions with fixed [3] W. Dong, X. Liu, C. Chen, Y. He, G. Chen, Y. Liu, and J.
number of retransmissions (ASn, where n ≥ 1 is the Bu, “DPLC: dynamic packet length control in wireless sensor
retransmission number), and unreliable transmissions (AS0). networks,” in Proc.2010 IEEE INFOCOM.

196
 IJISET - International Journal of Innovative Science, Engineering & Technology, Vol. 2 Issue 2, February 2015.
www.ijiset.com
ISSN 2348 – 7968
[4] B. Chen, Z. Zhou, Y. Zhao, and H. Yu, “Efficient error [22] Y. Sankarasubramaniam, I. F. Akyildiz, and S. W.
estimating coding: feasibility and applications,” in Proc. 2010 Mclaughlin, “Energy efficiency based packet size
ACM SIGCOMM. optimization in wireless sensor networks,” in Proc. 2003 IEEE
[5] J. Huang, G. Xing, G. Zhou, and R. Zhou, “Beyond co- International Workshop Sensor Netw. Protocols Applications.
existence: exploiting WiFi white space for ZigBee [23] I. F. Akyildiz, W. Su, Y. Sankarasubramaniam, and E.
performance assurance,” in Proc. 2010 IEEE ICNP. Cayirci, “Wireless sensor networks: a survey,” Comput.
[6] O. Gnawali, R. Fonseca, K. Jamieson, D. Moss, and P. Netw., vol. 38, pp. 393–422, 2002.
Levis, “Collection tree protocol,” in Proc. 2009 ACM SenSys. [24] E. Modiano, “An adaptive algorithm for optimizing the
[7] S. Ganeriwal, I. Tsigkogiannis, H. Shim, V. Tsiatsis, M. B. packet size used in wireless ARQ protocols,” Wireless Netw.,
Srivastava, and D. Ganesan, “Estimating clock uncertainty for vol. 5, no. 4, pp. 279–286, 1999.
efficient duty-cycling in sensor networks,” IEEE/ACM Trans. [25] P. Lettierri and M. B. Srivastava, “Adaptive frame length
Netw., vol. 17, no. 3, pp. 843–856, 2009. control for improving wireless link throughput, range, and
[8] P. R. Jelenkovi´c and J. Tan, “Dynamic packet energy efficiency,” in Proc. 1998 IEEE INFOCOM.
fragmentation for wireless channels with failures,” in Proc. [26] J. D. Spragins, J. L. Hammond, and K. Pawlikowski,
2008 ACM MobiHoc. Telecommunications: Protocols and Design. Addison Wesley
[9] G. Hackmann, O. Chipara, and C. Lu, “Robust topology Publishing Company,1991.
control for indoor wireless sensor networks,” in Proc. 2008 [27] C. K. Siew and D. J. Goodman, “Packet data transmission
ACM SenSys. over mobile radio channels,” IEEE Trans. Veh. Technol., vol.
[10] F. Zheng and J. Nelson, “Adaptive design for the packet 38, no. 2, pp. 95–101,1989.
length of IEEE 802.11n networks,” in Proc. 2008 IEEE ICC.
[11] K. Jamieson and H. Balakrishnan, “PPR: partial packet
recovery for wireless networks,” in Proc. 2007 ACM
SIGCOMM.
[12] S. Kim, R. Fonseca, P. Dutta, A. Tavakoli, D. Culler, P.
Levis, S. Shenker, and I. Stoica, “Flush: a reliable bulk
transport protocol for multihop wireless networks,” in Proc.
2007 ACM SenSys.
[13] L. Sang, A. Arora, and H. Zhang, “On exploiting
asymmetric wireless links via one-way estimation,” in Proc.
2007 ACM MobiHoc.
[14] X. Liu, Q. Wang, W. He, M. Caccamo, and L. Sha,
“Optimal real-time sampling rate assignment for wireless
sensor networks,” ACM Trans. Sensor Netw., vol. 2, no. 2, pp.
263–295, 2006.
[15] H. Zhang, A. Arora, and P. Sinha, “Learn on the fly: data-
driven link estimation and routing in sensor network
backbones,” in Proc. 2006 IEEE INFOCOM.
[16] K. Srinivasan and P. Levis, “RSSI is under appreciated,”
in Proc. 2006 EmNets.
[17] M. Buettner, G. Yee, E. Anderson, and R. Han, “X-MAC:
a short preamble MAC protocol for duty-cycled wireless
sensor networks,” in Proc. 2006 ACM SenSys.
[18] J. Polastre, R. Szewczyk, and D. Culler, “Telos: enabling
ultra-low power wireless research,” in Proc. 2005 ACM/IEEE
IPSN.
[19] H. Dubois-Ferrière, D. Estrin, and M. Vetterli, “Packet
combining in sensor networks,” in Proc. 2005 ACM SenSys.
[20] J. W. Hui and D. Culler, “The dynamic behavior of a data
dissemination protocol for network programming at scale,” in
Proc. 2004 ACM SenSys.
[21] T. He, B. M. Blum, J. A. Stankovic, and T. Abdelzaher,
“AIDA: adaptive application-independent data aggregation in
wireless sensor networks,” ACM Trans. Embedded Comput.
Syst., vol. 3, no. 2, pp. 426–457, 2004.

197