Resilient Distributed Maximum Correntropy Filter for Non-Gaussian Noise and Replay Attacks

This paper investigates the resilient distributed estimation problem in multi-sensor networks under non-Gaussian noises and replay attacks. Considering the advantage of correntropy in capturing the higher-order statistical characteristics, a new distributed maximum correntropy filtering algorithm within the framework of consensus on information is proposed. The algorithm utilizes the propagation of information pairs from local sensors to achieve effective filtering of non-Gaussian noises with limited communication resources. Then, a timestamp mechanism is integrated to enhance resilience against replay attacks by redesigning the correntropy-based weighting constant. Under the conditions of global observability and strongly connected topology, the consistency and boundedness of our proposed filter are proven theoretically, and an analytical assessment of replay attacks is established. Finally, a numerical simulation is presented to demonstrate the algorithm’s effectiveness.