Information Fusion and Target Tracking: Information‐Theoretic Sensor Selection
Nianxia Cao, Pramod K. Varshney, Engin Maşazade, Sora Haley · 2024
This chapter studies the problem of sensor selection for target tracking in a wireless sensor network from an information-theoretic perspective. To balance the computation cost and tracking performance, a sensor selection metric based on mutual information (MI) upper bound is proposed. The performance of the proposed metric is compared with a Fisher Information and a MI-based sensor selection metric. Fisher information-based sensor selection is simple but achieves lower computational complexity, while MI-based sensor selection provides better tracking accuracy but is computationally expensive. Furthermore, a multiobjective optimization framework is considered for information fusion to reveal trade-offs between the number of active sensors and tracking performance.