Acoustic Source Positioning based on Sensor Selection in Wireless Sensor Network

Abstract

Since the bandwidth and energy of wireless sensor networks (WSNs) are limited, it is not appropriate to use all the sensors for acoustic source positioning and so the need for sensor selection. In the article, an efficient expectation maximization algorithm based on sensor selection (EM-SS) is proposed for acoustic source positioning in the WSNs. The sensor selection solution based on the generalized information gain is introduced to select a subset of sensors which can provide reliable measurements. The information filter only depends on the Boolean decision variables and may make full use of the structure of measurement noise. Fewer sensor nodes are used and mass energy is economized. Simulation results demonstrate the well performance of the EM-SS algorithm in terms of localization accuracy, while only a part of sensors is used, so mass energy is economized and the communication channel is smooth.