Motion probes for fault detection and recovery in networked control systems

In this paper we address the problem of how to excite networked control systems in such a way that faulty, and possibly malicious, agents can be detected. In particular, we envision a scenario in which a subset of the agents in the network are executing a different control strategy than what has been prescribed and the problem under consideration consists of two subtasks, namely detection of the faulty agents, and isolation and mitigation of their impact on the remaining agents. We achieve this by proposing a collection of "motion probes" that leave certain aspects of the network invariant while ensuring that non-cooperative agents are identified and isolated. While we propose a useful tool to identify the faulty agents, the actual algorithms and policies that make use of the motion probes are left to future research.