Congestion control in wireless sensor networks

Yaakob, N 2013, Congestion control in wireless sensor networks, Doctor of Philosophy (PhD), Computer Science and Information Technology, RMIT University.


Document type: Thesis
Collection: Theses

Attached Files
Name Description MIMEType Size
Yaakob.pdf Thesis application/pdf 12.06MB
Title Congestion control in wireless sensor networks
Author(s) Yaakob, N
Year 2013
Abstract Information-sensing and data-forwarding in Wireless Sensor Networks (WSN) often incurs high traffic demands, especially during event detection and concurrent transmissions. Managing such large amounts of data remains a considerable challenge in resource-limited systems like WSN, which typically observe a many-to-one transmission model. The result is often a state of constant buffer-overload or congestion, preventing desirable performance to the extent of collapsing an entire network. The work herein seeks to circumvent congestion issues and its negative effects in WSN and derivative platforms such as Body Sensor Networks (BSN).

The recent proliferation of WSN has emphasized the need for high Quality-of-Service (QoS) in applications involving real-time and remote monitoring systems such as home automation, military surveillance, environmental hazard detection, as well as BSN-based healthcare and assisted-living systems. Nevertheless, nodes in WSN are often resource-starved as data converges and cause congestion at critical points in such networks. Although this has been a primal concern within the WSN field, elementary issues such as fairness and reliability that directly relate to congestion are still under-served. Moreover, hindering loss of important packets, and the need to avoid packet entrapment in certain network areas remain salient avenues of research. Such issues provide the motivation for this thesis, which lead to four research concerns: (i) reduction of high-traffic volumes; (ii) optimization of selective packet discarding; (iii) avoidance of infected areas; and (iv) collision avoidance with packet-size optimization. Addressing these areas would provide for high QoS levels, and pave the way for seamless transmissions in WSN.

Accordingly, the first chapter attempts to reduce the amount of network traffic during simultaneous data transmissions, using a rate-limiting technique known as Relaxation Theory (RT). The goal is for substantial reductions in otherwise large data-streams that cause buffer overflows. Experimentation and analysis with Network Simulator 2 (NS-2), show substantial improvement in performance, leading to our belief that RT-MMF can cope with high incoming traffic scenarios and thus, avoid congestion issues.

Whilst limiting congestion is a primary objective, this thesis keenly addresses subsequent issues, especially in worst-case scenarios where congestion is inevitable. The second research question aims at minimizing the loss of important packets crucial to data interpretation at end-systems. This is achieved using the integration of selective packet discarding and Multi-Objective Optimization (MOO) function, contributing to the effective resource-usage and optimized system.

A scheme was also developed to detour packet transmissions when nodes become infected. Extensive evaluations demonstrate that incoming packets are successfully delivered to their destinations despite the presence of infected nodes. The final research question addresses packet collisions in a shared wireless medium using distributed collision control that takes packet sizes into consideration. Performance evaluation and analysis reveals desirable performance that are resulted from a strong consideration of packet sizes, and the effect of different Bit Error Rates (BERs).


Degree Doctor of Philosophy (PhD)
Institution RMIT University
School, Department or Centre Computer Science and Information Technology
Keyword(s) congestion control
wireless sensor networks
multi-objectives optimization
selective packet discarding
relaxation theory
by-passed routing
collision avoidance
Versions
Version Filter Type
Access Statistics: 467 Abstract Views, 1116 File Downloads  -  Detailed Statistics
Created: Fri, 22 Aug 2014, 11:37:31 EST by Denise Paciocco
© 2014 RMIT Research Repository • Powered by Fez SoftwareContact us