Sensor scheduling for target tracking: A Monte Carlo sampling approach

We study the problem of sensor-scheduling for target tracking—to determine which sensors to activate over time to trade off tracking performance with sensor usage costs. We approach this problem by formulating it as a partially observable Markov decision process (POMDP), and develop a Monte Carlo solution method using a combination of particle filtering for belief-state estimation and sampling-based Q-value approximation for lookahead. To evaluate the effectiveness of our approach, we consider a simple sensor-scheduling problem involving multiple sensors for tracking a single target.