AbstractsComputer Science

Wireless sensor networks (WSNs) are designed for sensing phenomena and acquiring data. In structural health monitoring (SHM) of engineering structures, increasingly large number of sensor nodes are deployed to acquire data at the spatial density, needed for structural integrity assessment. During catastrophic events like earthquake there is a surge in simultaneous production and transmission of data to a central server at remote location. The increased contention for the wireless channel increases the probability of packet collisions resulting in packet drops, multiple transmission attempts and consequent delays. It is also not uncommon to find certain nodes (e.g. closer to sink) having better success rate in transmission of data and thereby leading to biased data delivery. Many solutions to the problem exist and clustering is the most commonly used method among then, wherein sensor nodes are grouped together. While the existing clustering algorithms do solve the network contention problems, the problem of cluster bias induced due to the proximity to sink node still remains to be addressed. Moreover all the existing solutions are very much node centric. This thesis presents a new perspective on cluster based WSNs designed to tackle Medium Access Control (MAC) layer congestion associated with burst packet generation in an unbiased manner, thereby making it more efficient for applications like SHM. In addition to solving the network bias problem, the proposed design also ensures faster transmission times, increased throughput and energy efficiency. Advisors/Committee Members: Seah, Winston.