Distributed maximum correntropy filtering for a class of multi-rate systems over sensor networks

Abstract

In this paper, the distributed maximum correntropy filtering problem is investigated for a class of multi-rate systems over sensor networks. The system under consideration is monitored by a sensor network whose sensors are permitted to have different sampling periods. The maximum correntropy criterion is employed to handle the non-Gaussian noises effectively. Given the distributed nature and the asynchronous sampling of the sensor network, a novel filtering performance index based on the correntropy is constructed for each node in the sensor network. To maximize this correntropy-based index, the desired filter gain at each time step is calculated using a fixed-point algorithm. Sufficient conditions for ensuring the convergence of the fixed-point method are established. Finally, the efficacy of the proposed filtering scheme is demonstrated through a target tracking example.