A saturated dynamic input allocation policy for preventing undetectable attacks in cyber-physical systems

The design of a dynamic allocator is proposed to tackle the presence of undetectable attacks in cyber-physical systems. Generating invisible signals that are capable to maintain the control inputs proximal to the saturation limits, the attacks are naturally discovered due to constraint violation. Two different design schemes are proposed: the first one is a direct allocation method based on a simple linear programming algorithm, while the second one is a dynamic optimization problem with a barrier function. The case-study of a consensus network illustrates the applicability and the efficacy of the proposed policy.