Cooperative Target Tracking Using Decentralized Particle Filtering and RSS Sensors

This paper introduces new cooperative particle filter algorithms for tracking emitters using received-signal strength (RSS) measurements. In the studied scenario, multiple RSS sensors passively observe different attenuated and noisy versions of the same signal originating from a moving emitter and cooperate to estimate the emitter state. Assuming unknown sensor noise variances, we derive an exact decentralized implementation of the centralized particle filter solution for this problem in a fully connected network. Next, assuming only local internode communication, we introduce two fully distributed consensus-based solutions to the cooperative tracking problem using respectively average consensus iterations and a novel ordered minimum consensus approach. In the latter case, we are able to reproduce the exact centralized solution in a finite number of consensus iterations. To further reduce the communication cost, we derive in the sequel a new suboptimal algorithm which employs suitable parametric approximations to summarize messages that are broadcast over the network. Numerical simulations with small-scale networks show that the proposed approximation leads to a modest degradation in performance, but with much lower communication overhead. Finally, we introduce a second alternative low communication cost algorithm based on random information dissemination.