Joint object detection and tracking in sensor networks

A nonlinear filtering based approach that fuses sensor data from the local sensors is proposed to jointly detect and track a moving object in a sensor field. First, the optimal detection algorithm based on the optimal nonlinear filter and the likelihood ratio test is provided. Then, a computationally efficient approach based on the extended Kalman filter is proposed and applied to jointly detect and track an object with very weak signal in a passive sensor network. The signal intensity is assumed to be inversely proportional to a power of the distance from the object. Simulation results show that the proposed detection approach can quickly detect the object after it appears in the sensor field with very high detection performance, even when the object state estimate is not very accurate.