Exploiting Interlaced Kalman Filter for Anomaly Detection

This paper introduces a strategy for detecting integrity attacks in cyber-physical systems, like false data injection, in a distributed manner. The core of the proposed method lies in employing an Interlaced Kalman Filter. Each filter collects localized measurements related to the system’s inputs and outputs, allowing for concurrent estimation of the overall system state. The exchange of state estimate and covariance matrices between neighboring filters enhances the precision of estimation. The algorithm has been enhanced to compensate for estimation uncertainty when subsystems cannot synchronize, with the aim of maximizing the consistency and content of the information.Each filter incorporates a detection mechanism to identify measurement anomalies and prevent inaccurate estimations. Leveraging the central limit theorem, a batch analysis ensures higher estimation accuracy and consistency of the results.Contributions of the paper include balancing computational burden and algorithmic complexity while addressing integrity attacks on sensor readings. The efficacy of the proposed algorithm is validated through a scenario involving nine water tanks grouped into four subsystems.