Control Theoretic Sensor Deployment Approach for Data Fusion Based Detection

Abstract

In this paper, we study the sensor deployment problem in a value fusion based distributed sensor network (DSN) detection system. More specifically, we study the problem of determining the positions at which a fixed number of sensors can be deployed in order to minimize the squared error (SE) between achieved and required detection probabilities while satisfying false alarm requirements. We show that this deployment problem can be modeled as a linear quadratic regulator problem (LQR). Subsequently, we develop two deployment algorithms; an optimal control based and a suboptimal deployment algorithm. We compare the performance of the proposed algorithms to that of a greedy deployment algorithm. Results indicate that the proposed algorithms have a faster SE convergence rate than that of the greedy algorithm. As a result, the proposed algorithms can use as much as 25% fewer number of sensors than the greedy algorithm to satisfy the same detection and false alarm requirements.