Direct Tracking of a Wireless Transmitter Based on Rao–Blackwellized Point Mass Filter

In this article, we address the problem of direct tracking of a wireless transmitter. That is, the inputs given to the Bayesian filter are the received baseband signals instead of pre-computed ranges or angles. We first propose to use the Rao-Blackwellized Point Mass Filter (RBPMF) to solve such a tracking problem. As such, the resulting tracking solution is still computationally expensive. Therefore, we propose an approach for reducing the computational cost of the RBPMF. More precisely, we replace the prediction step by the one of the Linear Kalman Filter (LKF). This combination helps to avoid expensive operations such as the weight convolution in the prediction step. In addition, it also allows complexity reductions in the correction step. As a result, the complexity is reduced by one order of magnitude compared to the original RBPMF. We compare our approach to representative direct-tracking methods, based on Iterative Extended Kalman Filter (IEKF) and Particle Filter (PF). The proposed solution has lower and comparable localization error compared to IEKF and PF, respectively. In addition, the proposed solution is of slightly less complexity than PF. However, the complexity reduction is significant compared to the conventional RBPMF.