Dynamic Agents Self-Deployment for Tracking Moving Targets Based on Target Motion Models

Tracking multiple navigating targets in a bounded region is a common problem that arises in many real-life applications, such as surveillance and reconnaissance. This paper introduces a distributed approach for a well-defined tracking problem, which can be shown to be an NP-hard problem. First, a stochastic target motion model that can be used by agents for dynamic deployment is introduced. Given the motion model, agents predict target location probabilities and compute their next near-optimal locations based on the predictions. The proposed approach involves coordination between mobile agents in order to achieve a near-optimal global utility. The approach has been evaluated through a number of simulation experiments. Simulation results reveal the superiority of the proposed model-based approach over existing approaches. Key-Words: Sensor Placement, Decentralized Algorithms, Target Tracking, Stochastic Motion Models