Investigations on distributed detection in wireless sensor networks

Abstract

Wireless sensor networks formed by numerous tiny devices capable of sensing computing and wireless communication are emerging as a revolutionary technology with applications in diverse areas The unique features of wireless sensor networks have brought new challenges and problems to the field of distributed and collaborative newlineinformation processing In this dissertation some important problems within the broad topic of distributed detection are addressed with an emphasis on the efficient use of stringent system resources to achieve certain application specific objectives In this thesis important results on distributed detection are presented considering the parallel architecture and decision rules obtained from the Neyman Pearson criterion General comments on several important issues such as sensor fault measurement noise channel error and source signal characteristics are made The optimal design of WSNs using the Neyman Pearson methodology for monitoring and detecting the possible presence of ionizing radiations, in the vicinity of a nuclear power plant is considered as a newlinerepresentative design example The design equations for the wireless sensor network with parallel configuration focusing only on the signal processing task under certain assumptions are derived The detection performance of the individual node and network of the sensor nodes under two different operating options for different network parameters and sensor characteristics are studied to understand the design trade offs between the sensor network parameters and the performance measures newline