摘要：AbstractIn this paper, we consider the maximum-value based approach towards the resilient consensus problem. The network consists of agents with unknown identities having faults or malicious intentions. Their objective is to prevent the nonfaulty, normal agents from reaching consensus. We extend our recent results on resilient versions of consensus algorithms where at updates, the agents ignore some of the neighbor values to avoid being influenced by the malicious agents. In particular, the normal agents attempt to converge at the maximum value of their initial states. The main advantage of this maximum-based approach is the fast convergence which is shown to be finite. We deal with both the synchronous and asynchronous update schemes. In our analysis, we characterize the graph theoretic conditions for the proposed algorithm to achieve resilient consensus. The advantages of the approach are illustrated by means of a numerical example.
