Model Based Information Fusion in Sensor Networks

In this paper, a model based sensor fusion algorithm for sensor networks is presented. The algorithm, referred to as distributed Kalman filtering is based on a previously presented algorithm with the same name. The weight selection process has been improved yielding performance improvements of several times for the examples studied. Also, solutions to both optimization problems involved in the iterative off-line weight selection process are given as closed form expressions. The algorithm is also demonstrated on a typical signal tracking application