Binary Observation-Based FIR System Identification Against Replay Attacks

Abstract

In the context of the increasing security issues of cyber-physical systems (CPSs), this paper addresses the parameter identification of binary observation-based finite impulse response (FIR) systems under replay attacks, overcoming the problem of high nonlinearity of quantized systems and greater data sparsity caused by replay attacks. For the attacker, based on the energy-constrained condition, an optimization attack model is established to maximize the absolute error of identification, giving the method of obtaining the optimal attack strategy. Following the defender, the identifiability of unknown parameters is discussed and a robust defense scheme is proposed. This scheme involves a joint identification strategy for both the attack strategy and unknown parameters. By enhancing the excitability of system inputs, consistent identification is ensured despite replay attacks. An algorithm for the accomplishment of the joint identification strategy is presented based on the grid search method. Rationality of the method is confirmed with performing numerical simulations.