Resilient State Estimation for Complex Networks With Partial Node Measurements and State Saturations: Defense Against Replay Attacks

This paper addresses the problem of resilient state estimation for complex networks, with a particular focus on scenarios involving state saturations and replay attacks, using a partial-nodes-based approach. State saturation, characterized as a type of nonlinearity, is considered to reflect practical engineering scenarios. Replay attacks, executed by adversaries on communication channels between sensors and estimators, are defined by the replacement of current measurements with previously recorded measurements. The dynamics of these replay attacks are described by two components: one dependent on a stochastic variable and the other on a time-varying parameter. To mitigate the adverse effects of state saturations and replay attacks on estimation performance, a partial-nodes-based resilient estimator is designed. By employing the convex hull method, a sufficient condition is derived to ensure that the augmented error system is exponentially ultimately bounded in the mean-square sense. Furthermore, the gain parameter of the estimator is determined by solving a specific matrix inequality. Finally, the feasibility and effectiveness of the proposed estimation approach are validated through numerical simulations.