Simulative Study of Two Fusion Methods for Target Tracking in Wireless Sensor Networks

Abstract

Target tracking is one of the popular applications of wireless sensor networks wherein hundreds or thousands of randomly distributed sensor nodes in an environment gather spatio-temporal information from target(s) and send them to a sink node for further processing. Due to various environmental factors on sensor devices, this information is seldom very accurate. Sensor nodes partly process their sensed data using local fusion before sending them to the sink. This paper comparatively studies two major voting algorithms for fusion of target tracking data in intermediate nodes with a view on the accuracy of results. Majority voter and mean voter algorithms are simulated with different densities of sensor nodes to determine the best choice of sensor density for cost effective deployment of sensor nodes. It is shown that formal majority voter yields much more accurate and stable results in location tracking applications than mean voter.