DISTRIBUTED INFORMATION FUSION WITH INTERMITTENT OBSERVATIONS FOR LARGE-SCALE SENSOR NETWORKS

Abstract

In this paper, we present a robust distributed fusion algorithm to handle intermittent observations via an interacting multiple model (IMM) and a sliding window strategy which is applied to large-scale sensor networks. Intermittent observations are frequently occurred in practice especially when the scale of network becomes larger and sensors are dynamically connected. To solve the problem, we model the communication channel as a jump Markov system and a posterior probability distribution of communication channel characteristics is calculated and incorporated into the filter. By doing so, the distributed Kalman filtering can automatically handle the intermittent observation situations. For the implementation of the distributed fusion, a Kalman-Consensus filter (KCF) is adopted to provide the average consensus based on the estimates of distributed sensors over a large-scale sensor network. In addition, the algorithm is extended to nonlinear systems so as to be implemented for more general dynamic systems. The advantages of proposed algorithm are subsequently verified from target tracking examples for a lame-scale network with intermittent observations.