Target tracking in collaborative sensor networks by using a decentralized leader-based scheme

In this paper, the problem of fully-decentralized data fusion is addressed for tracking a target with nonlinear motion model. The problem is solved by applying a fully-decentralized estimation algorithm based on the extended information filter. We propose the neighbor selection and information selection algorithms for sensor selection based on their closeness to the estimated target position and their information contribution, respectively. Simulation results show that compared to the DEIF algorithm we obtain an analogous response with less consumption of energy and computations.