Distributed optimal H2/H∞ filtering over unreliable wireless sensor networks

This paper investigates the problem of distributed H-2/H-infinity filtering through unreliable wireless sensor networks, where data packet drops occur during the signal transmission from neighbor nodes to local nodes due to limited communication resources. To optimize estimation performance in the H-2 and H-infinity senses, a matrix decomposition method is proposed to settle the couplings resulting from data packet drops. Compared with the existing H-infinity performance optimization methods, a less-conservative design against the non-zero initial condition is derived by introducing a modified cost function involving two H-infinity performance indexes. Moreover, in the distributed framework with data drops, the estimation performance of the local node is related to data arrival rates of relevant channels and, by proposing a notion of transmission capacity referring to data arrival rate, the H-2 performance margin of any one channel's transmission capacity is obtained. Especially, based on an equivalent form of a modified algebraic Riccati equation (MARE), a necessary and sufficient condition is further established for the existence of a local filter satisfying an expected H-2 performance. A simulation of a radar tracking system is given to demonstrate the effectiveness of the proposed methods.