Robustness of Attack-Resilient State Estimators

摘要: The interaction between information technology and physical world makes Cyber-Physical Systems (CPS) vulnerable to malicious attacks beyond the standard cyber attacks. This has motivated need for attack-resilient state estimation. Yet, existing state-estimators are based on non-realistic assumption that exact system model is known. Consequently, in this work we present a method estimation presence of attacks, systems with noise modeling errors. When estimated states used by state-based feedback controller, show attacker cannot destabilize exploiting difference real dynamics system. Furthermore, describe how implementation issues such as jitter, latency synchronization errors can be mapped into parameters procedure errors, provide bound state-estimation error caused enables mapping control performance requirements real-time (i.e., timing related) specifications imposed underlying platform. Finally, illustrate experimentally evaluate approach an unmanned ground vehicle case-study.