Information Weighted Consensus With Interacting Multiple Model Over Distributed Networks

Abstract

Distributed estimation approach is becoming increasingly popular in the sensor networks community. In this brief, an information weighted consensus with interacting multiple models is proposed for distributed networks. Firstly, the multiple models predict the state estimate individually at each node. Then, the measured data across nodes are fused effectively through the local communication among neighboring nodes. Afterward, the fused data are employed to update the state estimates predicted by multiple models at each node. Finally, a novel model probability calculation criterion is presented to obtain the global state estimate at each node. The effectiveness of the proposed method is demonstrated on a target tracking task.