Information-weighted Consensus Filtering under Limited Communication

Abstract

A distributed sensor network refers to a set of stationary or mobile agents equipped with sensors, each of which observes and estimates the state of some system, while utilizing the communication channels of the sensor network to improve the overall state estimation performance of the network. In particular, consensus-based distributed state estimation algorithms can compute accurate state estimates of the system at each agent under relaxed observability conditions, using a low communication bandwidth. A drawback of many existing consensus-based approaches is that they are developed under the presumption that a large number of consensus subiterations are used, which may not be possible in large-scale sensor networks, in the presence of communication delays and/or energy constraints. In this paper, we develop an extension of the Information-weighted Consensus Filter (ICF) for the case where only a limited number of consensus subiterations are possible. Through mathematical analysis and numerical simulations, it is shown that the proposed algorithm improves upon the existing algorithms under identical computation and communication requirements.