Towards General Performance Bounds of the Distributed Sensor Coverage Problem

Graduate Thesis Or Dissertation

Citations

Citeable URL: https://scholar.colorado.edu/concern/graduate_thesis_or_dissertations/5m60qs283

Abstract

We study a class of resource allocation problems know as distributed sensor coverage problems, whereby sensors are distributed to regions in space with the goal of maximizing the detection high value events. Computing an optimized allocation of sensors onto regions would become intractable for a centralized controller in large scale system and would require continuous communication to all agents in the system, which may not be feasible. We analyze the alternate approach of a distributed sensor coverage problem, where each sensor is responsible for selecting a region to search. When each sensor acts in their own interests to select a resource, degradation in system performance can occur. Game Theory is used as a mathematical framework to model and study the behavior of large groups of agents in the sensor coverage problem. We study the lower bound on performance of the distributed sensor coverage problem and conjecture that this bound is e/(e+1). We support this claim with provable properties about the structure of games bounded by this price of anarchy, and through empirical simulations.

Creator