Correlated movement mobility model and constant acceleration model for EKF-based tracking applications

The extended Kalman filter (EKF) is an important tool in wireless sensor networks for localization and tracking of a dynamic target. The performance of the EKF is heavily dependent on the mathematical model used for target dynamics. Significant research has been done for finding an optimum state estimation model for target motion. In this paper we propose a new mobility model for target movement simulations. The correlated movement (CM) mobility model is presented for simulating of the natural movements of a real life target. It is shown to be more realistic for the purpose of performance testing, and it facilitates better comparisons of tracking algorithms. The use of this mobility model is demonstrated in conjunction with EKF-based target tracking algorithms based on constant velocity (CV) and constant acceleration (CA) assumptions. For larger fluctuations in the speed of a target, simulation results indicate that the performance of the CA algorithm is superior to that of the CV algorithm in terms of better tracking quality and reduced failure rates. The CA algorithm also allows for larger time intervals to be used in between distance measurements, thus increasing the efficiency of energy utilization in the wireless sensor network.