Distributed information fusion filter with intermittent observations

Abstract

We present a robust distributed fusion algorithm with intermittent observations via an interacting multiple model (IMM) approach and sliding window strategy that can be applied to a large-scale sensor network. The communication channel is modelled as a jump Markov system and a posterior probability distribution for communication channel characteristics is calculated and incorporated into the filter to allow distributed Kalman filtering to automatically handle the intermittent observation situations. To implement distributed Kalman filtering, a Kalman-Consensus filter (KCF) is then used to obtain the average consensus based on the estimates of distributed sensors over a large-scale sensor network. From a target-tracking example for a large-scale sensor network with intermittent observations, the advantages of proposed algorithms are subsequently verified.