Decentralized-Detection Based Mobile Multi-Target Tracking in Wireless Sensor Networks

Due to the communications and energy constraints, we propose the decentralized-detection based schemes for Mobile Multi-Target Tracking (MMTT) in Wireless Sensor Networks (WSN). Our developed WSN consists of a symmetric-tree structure and a set of target detection and estimation strategies, which achieve the optimal error-exponent decay in detecting the number of the tracked targets. The increase in the target-number estimate accuracy also yields the high target-position estimate accuracy. We apply the Decentralized Probability Hypothesis Density (DPHD) filtering algorithm in deriving the global optimal threshold-levels for our proposed strategies to maximize the estimating accuracy for the positions of the tracked targets. The obtained extensive evaluation analyses validate and evaluate our proposed decentralized-detection structure of WSN and our developed target tracking strategies.