Multiple Target Tracking in Wireless Networks Based on Time-frequency Distributions

We present a new method for multitarget tracking in wireless sensor networks (WSNs) based on chirplet decomposition. In all sensors of the network, reflected signals from targets are decomposed into chirplets. Through this new formulation, after clutter rejection, the delay and Doppler are expressed in terms of chirplet parameters. As an advantage of this strategy, the background noise may assume an arbitrary distribution. Each pair of delay and Doppler is associated with a target based on minimum distance from all pairs of previous scans. Finally, estimated delay and Doppler pairs are utilized to track all targets by an extended Kalman filter (EKF).