Trust-aware filtering in the presence of non-Gaussian noise and cyber attacks

Abstract

Cyber attacks and non-Gaussian noise interference present unique security challenges to Wireless Sensor Networks (WSNs). Despite the existence of many techniques to resist non-Gaussian noise and attacks, when the system is simultaneously affected by both, non-Gaussian noise can, to some extent, affect attack detection and mitigation, leading to the failure of attack detection and degradation of system performance. In this paper, we propose a trust-based distributed Kalman filtering technique. We introduce a new trust evaluation metric combined with clustering methods for identifying attacked nodes. The maximum correntropy Kalman filter (MCKF) is employed for information fusion to mitigate the effects of non-Gaussian noise. Additionally, a malicious detection mechanism based on trust metrics' similarity is proposed. Compared to recently proposed trust-based methods, simulation results demonstrate that the proposed filter can simultaneously resist non-Gaussian noise interference and cyber attacks, with better performance.