An optimal graph-search method for secure state estimation

The growing complexity of modern Cyber–Physical Systems (CPS) and the frequent communication between their components make them vulnerable to malicious attacks. As a result, secure state estimation is critical requirement for control these systems. Many existing methods suffer from combinatorial which grows with number states sensors in system. This can be mitigated using optimization-based that relax original problem, although at cost optimality as often identify attack-free attacked. In this paper, we propose new optimal graph-search algorithm correctly attacks securely estimate even large-scale CPS modeled linear time-invariant systems graph consists layers, each one containing two nodes capturing truth assignment any given sensor, directed edges connecting adjacent layers only. Then, our searches incrementally, favoring directions higher more assignments, while actively managing repository expanded later iterations. proposed search bias ability revisit self-correct, allow reach faster tackle larger problems. We show complete provided process measurement noises do not dominate attack signal. Moreover, provide numerical simulations demonstrate attacked reconstruct state. Our method outperforms algorithms both terms execution time.