Alternative Actuation Paths for Ship Applications in the Presence of Cyber-Attacks

Safety-critical Cyber-Physical Systems (CPS) always need to maintain operational normalcy, which may include reconfiguring the control system in response to cyber threats. This paper studies three different approaches to determine when redundancies in control and actuation exist and identify alternative actuation paths that keep the system operating safely in the event of an attack to the primary actuation path. The first two approaches use classical control tools in system identification to determine potential inputs based on the desired force traces, while the third employs a supervised learning algorithm. The three methods are then experimentally compared on a small-scale naval ship testbed. The testbed, although compact, allows the easy and fast evaluation of different control routines for CPS. Laboratory experimental results, when the rudder has been attacked, demonstrate the accuracy of all three approaches in determining correct alternative actuation paths and comparisons among the strategies are drawn.