Random-based Hidden Moving Target Defense against Alert False Data Injection Attackers

Moving target defense (MTD) is a proactive defense strategy that change reactance of transmission lines to resist false data injection (FDI) attacks against state estimation in power systems. A hidden MTD (HMTD) is a superior MTD method, as it is stealthy to alert attackers. However, it is necessary to model more alert attacker types with strong capabilities to evaluate the hiddenness of the HMTD. This paper summarizes two alert attacker models and proposes a novel alert attacker model: bad-data-detection-based alert attackers that use chi-square detector to detect MTD; data-driven alert attackers that use unsupervised learning methods to detect MTD; and novel model-based alert attackers that use MTD operation models to estimate the dispatched line reactance and then use chi-square detector to verify the estimated values. This paper proposes a novel random-based HMTD (RHMTD) operation method, which utilizes random weights to introduce uncertainties and uses the derived hiddenness operation condition as constraints. The maximized line impedance changes enable MTD to detect FDI attacks without any delay, and the uncertainties of RHMTD make MTD effective to detect FDI attacks by the model-based alert attackers. Simulation results on the IEEE 14-bus systems show the efficacy of the proposed operation approach.