Skip to main content

Dynamic Diffusion for Congestion Avoidance in Wireless Sensor Networks

Sri Divya Deenadayalan’s Thesis Defense

June 15, 2012
10:00am – 12:30pm
MEC 301

Advisor

Dr. Murali Medidi

ABSTRACT

Wireless Sensor Networks (WSNs) are employed for either continuous monitoring or event-detection in the area of interest. In event-driven applications, it is critical to report the detected events in the area, and with sudden bursts of traffic possible due to spatially-correlated events or multiple events, the data loss due to congestion will result in information loss or delayed arrival of the sensed information. Congestion control techniques were proposed to detect congestion and attempt to recover from packet losses due to congestion, but they cannot eliminate or prevent the occurrence of congestion. Congestion avoidance techniques employ proactive measures to alleviate future congestion using parameters like queue length, hop count, channel conditions and priority index. However, maintaining and processing such information becomes a significant overhead for the sensor nodes and degrade the performance of the network. We propose a congestion avoidance MAC protocol which uses sensor nodes’ queue buffer length to estimate the congestion and diffuse traffic to provide a congestion-free routing path towards the base station. Event reporting, packet delivery ratio, is improved by dynamically diffusing the traffic in the network using multiple forwarders in addition to backup forwarding. We used the standard Network Simulator (NS2) to evaluate the performance of our protocol. Results show that our protocol has significant improvement in event reporting in terms of packet delivery ratio, throughput and delay compared to the baseline approach.