Command-governor-based stealthy attack design for estimation difference regulation

Abstract

In the field of cyber–physical systems, significant attention has been received by the problem of stealthy attack strategy design for remote estimators, for which most existing approaches are exploited by maximizing the difference between the healthy and attacked estimations. However, in some practical scenarios, it is more effective to regulate such difference to some specific set so as to achieve some desired damages, e.g., collisions of unmanned aerial vehicles. Motivated by this, in this paper a novel stealthy attack strategy is developed with the attack purpose of regulating, instead of maximizing, the estimation difference. By constructing the stealthy constraint and identifying the feasible reference set, it is shown that such a design problem can be transformed into that of regulation subject to constraints on estimation and innovation difference. Then, an on-line attack sequence design approach is proposed by incorporating the command governor into the linear quadratic regulator, which regulates the estimation difference to the target reference for achieving the desired attack purpose, while maintaining stealthiness. Numerical simulations are conducted to verify the effectiveness of the proposed attack strategy.