Bearing-only maneuvering mobile tracking with nonlinear filtering algorithms in wireless sensor networks

Mobile node localization is important to offer wireless services in vehicular communication applications. Some typical methods realize the mobile node tracking through data fusion from time of arrival (TOA) and received signal strength (RSS) measurements provided by sensor nodes or base stations (BSs). Although the TOA/RSS method is not expensive under a concern of cost, it is very sensitive to multipath signal propagation effects. As the technology of angle of arrival (AOA) antennas is showing a rapid progress, we turn to consider AOA estimation. In this paper, the nonlinear extended Kalman filter (EKF) and the particle filter (PF) along with a three-model interacting multiple model (IMM) algorithm are utilized and compared for maneuvering mobile station (MS) tracking with bearing-only measurements. A coordinated turn model is used to improve the tracking performance since the MS frequently turns in the streets. We also propose an efficient method for resampling particles to alleviate the degeneracy effect of particle propagation in the interacting multiple model particle filter (IMMPF) algorithm. Moreover, a BS sensor selection scheme is also exploited for the long-haul MS tracking case which often changes BSs in a wireless vehicular sensor network. Numerical simulations show that the three-model IMMPF algorithm outperforms the interacting multiple model extended Kalman filter algorithm and achieves a root-mean-square tracking performance which is quite close to the posterior Cramer-Rao lower bound.