Tracking of wireless mobile nodes in he presence of unknown path-loss characteristics

Due to the difficult characterization of the propagation model, most studies on racking of mobile nodes assume the correct knowledge of the power-distance gradients or the path-loss exponents (PLEs). In this paper, we first investigate the impact of erroneous PLEs on positioning of a wireless nodes when both distance and bearing measurements are available. Thus, an analytical expression of the mean square error (MSE) in location estimation is derived in case of erroneous PLEs. Second, we propose a novel online PLE estimation and tracking algorithm in dynamic environments. The proposed algorithm estimates the PLE of individual links a every time-step using he generalized pattern search (GenPS) algorithm. The PLE estimates update the observation vector which is used in a Kalman filter (KF) and a particle filter (PF) for tracking. Simulation results show that the racking performance degrades drastically with an incorrect assumption for the PLE values. Further simulations show that tracking with PLE estimation performs considerably beer compared to tracking with incorrectly assumed PLEs.