Object Tracking in Random Access Sensor Networks: Extended Kalman Filtering with State Overlapping

In this paper, we address the problem of object tracking using sensor networks where the sensor nodes measure the strength of the field generated by a number of objects, and transmit their measurements to a fusion center in a random access manner for final reconstruction of the trajectories. Our focus is on underwater systems that use acoustic communication. Extended Kalman filtering is employed for detection and tracking of the objects inside the observation area. We propose a method for object tracking called state overlapping, which is based on exchanging and overwriting the estimated state vector between a number of independent Kalman filters. The method improves the scalability of the system, relieves the requirement for a time-varying state vector, and reduces the probability of divergence. Moreover, we propose an adaptive rate control scheme and refine an existing one to improve the estimation accuracy and the energy efficiency of the system. The performance of these methods is evaluated through simulation, showing the effectiveness of the approaches proposed.