False noise attack detection for differentially-private distributed control of microgrids (Q6947649)

In this paper, the authors investigate the internal dynamic process in distributed energy management systems (DEMSs) with the differential privacy (DP) preserving method (DP-DEMSs). They propose a novel peer-to-peer (P2P) false noise (FN) (instead of normal noise in privacy-preserving algorithms) attack detection method, named false noise attack detection (FNAD). The proposed detection algorithm consists of three phases. In the first phase, each neighbor of a device individually observes the power decisions of the target device based on data from its two-hop neighbors. In the second phase, using the observation results, each neighbor estimates the power decision by the Kalman filter. In the last phase, each neighbor evaluates the detection index of the target device using the parameter estimation results at each iteration, and all neighbors vote to decide if a device is malicious when the detection index is above a well-defined threshold. The detection index is designed based on information entropy, while FNAD directly processes the discrete-time residue sequence without prior knowledge of the FN attack. The authors analyze the advantages of FNAD over traditional detection methods, noting that FNAD can effectively detect FN attacks without prior knowledge and preserve the privacy of each device. For more details see [A. S. Musleh, G. Chen and Z. Y. Dong, ``A Survey on the Detection Algorithms for False Data Injection Attacks in Smart Grids'', in IEEE Transactions on Smart Grid 11, No. 3, 2218--2234 (2020), doi: 10.1109/TSG.2019.2949998].