Covariance Intersection-Based Kalman Consensus Filtering with Sequential Replay Attack Detection Over Sensor Networks

This paper focuses on the security issue of consensus-based distributed filtering under replay attacks over wireless sensor networks. The local information from each sensor, which could be subject to replay attacks, is sent to its neighboring nodes through wireless communication channels. However, the existing strategies of replay attack detection cannot be migrated into the scenario where local estimates and error covariances are simultaneously transmitted. As for this concern, a new replay attack detection strategy is put forward to verify the received local estimate and error covariance for distributed filters consensus on information, respectively. Afterward, on the basis of the aforementioned strategy, a covariance intersection-based fusion algorithm is developed to fuse the local information sequentially. This algorithm guarantees the estimate consistency in the presence of replay attacks and unknown cross-correlations. Simultaneously, the stability properties of the proposed fusion algorithm with sequential attack detection strategy are analyzed. Finally, the effectiveness of the theoretical results is validated by two examples.