Date of Award

6-2012

Document Type

Thesis

Degree Name

Master of Science (MS)

Department

Computer Science

First Advisor

Dr. Hesham H. Ali

Second Advisor

Dr. Sanjukta Bhowmick

Third Advisor

Dr. Ken Dick

Abstract

While the majority of the current Wireless Sensor Networks (WSNs) research has prioritized either the coverage of the monitored area or the energy efficiency of the network, it is clear that their relationship must be further studied in order to find optimal solutions that balance the two factors. Higher degrees of redundancy can be attained by increasing the number of active sensors monitoring a given area which results in better performance. However, this in turn increases the energy being consumed. In our research, we focus on attaining a solution that considers several optimization parameters such as the percentage of coverage, quality of coverage and energy consumption. The problem is modeled using a bipartite graph and employs an evolutionary algorithm to handle the activation and deactivation of the sensors. An accelerated version of the algorithm is also presented; this algorithm attempts to cleverly mutate the string being considered after analyzing the desired output conditions and performs a calculated crossover depending on the fitness of the parent strings. This results in a quicker convergence and a considerable reduction in the search time for attaining the desired solutions. Proficient cluster formation in wireless sensor networks reduces the total energy consumed by the network and prolongs the life of the network. There are various clustering approaches proposed, depending on the application and the objective to be attained. There are situations in which sensors are randomly dispersed over the area to be monitored. In our research, we also propose a solution for such scenarios using heterogeneous networks where a network has to self-organize itself depending on the physical allocations of sensors, cluster heads etc. The problem is modeled using a multi-stage graph and employs combinatorial algorithms to determine which cluster head a particular sensor would report to and which sink node a cluster head would report to. The solution proposed provides flexibility so that it can be applied to any network irrespective of density of resources deployed in the network. Finally we try to analyze how the modification of the sequence of execution of the two methods modifies the results. We also attempt to diagnose the reasons responsible for it and conclude by highlighting the advantages of each of the sequence.

Comments

A Thesis Presented to the Department of Computer Science and the Faculty of the Graduate College University of Nebraska In Partial Fulfillment of the Requirements for the Degree Masters in Computer Science University of Nebraska at Omaha. Copyright 2012 Abhishek Karpate.

Download

COinS