Multi-Sensor Remote State Estimation under Critical-Data-Oriented Attacks over CPSs

This article investigates the security issue of malicious adversaries undermining the remote estimator over cyber-physical systems (CPSs). A critical-data-oriented (CDO) attack strategy is proposed to enable attackers to utilize essential data judiciously during data transmission, thereby degrading remote state estimation maximally. In addition, the proposed scheme is compared with commonly studied attack strategies such as Bernoulli-distributed and periodic denial-of-service (DoS) attacks within a unified framework. The evaluation of its impact focuses on two factors: the weighting matrix (WM) and the residual information. Furthermore, the consideration of attack energy distribution (AED) and quality of service (QoS) constraints is merged. The theoretical analysis based on stochastic methods reveals the inherent associations among WM, attack energy, and signal-to-interference-plus-noise ratio (SINR). The attack impact on channel throughput is analyzed and formulated as a convex optimization problem. The extended part explores the selection of optimal WM under special scenarios. Finally, two examples are established to validate the correctness and practicality of the concerned CDO attack strategy.