Diagnosis approaches for detection and isolation of cyber attacks and faults on a two-tank system

Recently, a two-tank benchmark for detection and isolation of cyber-attacks has been proposed to the diagnosis research community to test different diagnosis methods. In this work, we use this benchmark, add some scenarios, and test several diagnosis techniques to evaluate their diagnos-ability power. We propose to use a well-known model-based diagnosis approach to identify the groups of faults and attacks that are not isolable or detectable. In a second part, we investigate improvements provided by data-based diagnosis techniques and show that they succeed to isolate faults and attack for this benchmark