Posterior Crlb Based Sensor Selection for Target Tracking in Sensor Networks

The objective in sensor collaboration for target tracking is to dynamically select a subset of sensors over time to optimize tracking performance in terms of mean square error (MSE). In this paper, we apply the Monte Carlo method to compute the expected posterior Cramer-Rao lower bound (CRLB) in a nonlinear, possibly non-Gaussian, dynamic system. The joint recursive one-step-ahead CRLB on the state vector is introduced as the criterion for sensor selection. The proposed approach is validated by simulation results. In the experiments, a particle filter is used to track a single target moving according to a white noise acceleration model through a two-dimensional field where bearing-only sensors are randomly distributed. Simulation results demonstrate the improved tracking performance of the proposed method compared to other existing methods in terms of tracking accuracy.