Resilient consensus for multiagent networks with time-varying delay and mobile adversaries

Abstract Security issues in cyber-physical systems are often the Achilles heel for cooperative control and coordination. This paper addresses resilient consensus problems for multiagent systems under heterogeneous time-varying delays and mobile Byzantine adversaries. We propose two types of mobile attack models for continuous-time dynamical agents with or without infection detection capabilities. Delayed consensus protocols are presented where the agents are able to adjust their behavior during the latency period after recovery from an attack. It is shown that resilient consensus can be achieved under bounded delays and robust communication topologies. As a further contribution, resilient vector consensus has been solved within our framework. Simulations regarding the mobile patterns of adversaries indicate the irrelevance of global and local mobility in terms of convergence rate over Erdős–Rényi random networks. In contrast, local mobility of adversaries leads to longer convergence time compared to global mobility in power-law random networks due to their characteristic topology.