A Leader-Follower Attack-Tolerant Algorithm for Resilient Rendezvous with Reduced Network Redundancy

This paper addresses the resilient rendezvous problem for leader-follower multi-agent systems (MASs) in the presence of adversarial attacks. A novel leader-follower attack-tolerant (LFAT) algorithm is developed to ensure that the healthy followers reach rendezvous on the reference value propagated by healthy leaders. Compared with the existing weighted mean-subsequence-reduced (W-MSR) algorithm, the proposed LFAT algorithm includes a necessary state initialization step for leader and follower agents and an improved threat elimination step, so that more effective information can be retained for state updates. The necessary and sufficient condition on the network topology is further derived to ensure resilient rendezvous for leader-follower MASs. Compared with the existing resilient algorithms, the proposed LFAT algorithm enables MASs to achieve leader-follower resilient rendezvous under relaxed graph robustness conditions, so that the network redundancy is mitigated. Several numerical examples are given to illustrate the superior performance of the LFAT algorithm and the scalability to lager-scale and time-varying networks.