An approximate dynamic programming based non-myopic sensor selection method for target tracking

In this paper, we study the non-myopic sensor selection problem for target tracking in wireless sensor networks based on quantized sensor data. Using the conditional posterior Cramér-Rao lower bound (C-PCRLB) as a sensor selection metric, we formulate and solve a non-myopic sensor selection problem using an approximate dynamic programming (A-DP) algorithm. Given a constraint on the total number of selected sensors allowed while observing the target over a time window, simulation results show that the proposed non-myopic sensor selection scheme based on A-DP is computationally very efficient and yields better tracking performance than the myopic sensor selection scheme.