Sensor selection for target tracking in binary sensor networks using particle filter

In this paper, a new sensor selection scheme for target tracking in binary sensor networks using auxiliary particle filter is proposed that leads to computation complexity reduction. Binary sensor networks provide a set of binary signals based on presence or absence of target in detection region of sensors. Computation complexity of the particle filtering can be a major problem for its practical use in this application. We introduce a criterion for selecting neighbors for each sensor, and computation for target tracking is limited to one neighborhood of sensors, which is identified as the most important and effective neighborhood. Number of Floating-point Operations (FLOP) is considered as complexity measure in simulations, and accuracy of algorithms is investigated through Root Mean Square Error (RMSE). Simulation results show that our new method is less computationally complex in compare with an algorithm that is based on common auxiliary particle filter, and preserves the same accuracy.