Defending State-Feedback Based Controllers Against Sensor Attacks

Context: This paper is motivated by the need to improve resilience of industrial control systems. Many systems currently operating in industry were designed and implemented before boom communications (wired wireless networks) within However, nowadays they operate connected network. increase connectivity has made susceptible cyber-attacks that seek degrade proper operation loop even affecting only one sensor.
 Method: We use concepts from fault tolerant classic theory show it possible reconstruct system state without (any) outputs. fact used recalculation signal through an algorithm attack detection isolation, avoid for be feedback system, mitigating its effect. effectiveness our proposal with simulations on a four-tanks testbed using Matlab Simulink.
 Results: bank unknown input observers can recover true information attacked sensors, i.e., effect attack. Therefore, estimation obtained those utilized computing action mitigates attack.
 Conclusions: mitigation prevents single sensor signi cantly impair low-level controllers, improving just modifying digital controller architecture. The development presented limited sensors pre- sented at time, still compromise seriously behavior. Future work will address extension results situations simultaneous attacks more than sensor, and/or consider actuators.