Fusion State Estimation of CPS Under Quantized Observations and Tampering Attacks

With the development of industrialization, the security issue of remote state estimation in Cyber-Physical Systems (CPS) has attracted widespread attention in recent years. This paper studies the state estimation problem based on quantized observations and tampering attacks on partial channels of multi-sensor networks. Assuming the attacker knows the sending data and system parameters, we design an optimal attack strategy based on maximizing the state estimation error as an indicator. From the defender's perspective, when the attack strategy is known, we construct a compensation algorithm to reduce the error caused by the attack. When the attack strategy is unknown, we design an innovative attack strategy and system state joint estimation scheme. The defense scheme limits the attack strategy estimation error within a given range, while making the estimation performance of the system state almost the same as that without attack. The state estimators in the above attack and defense scenarios have been given and their performance is analyzed. Finally, the theoretical results are simulated and verified.