Resilient consensus of double-integrator multi-agent networks with communication delays

We consider a multi-agent network of second-order dynamics and study its resilient consensus taking account of cyber security issues. In our setting, some of the agents may be malicious whose aim is to prevent other normal agents from reaching consensus. They may do so by not following the given control laws and behaving arbitrarily. It is assumed that the total number of such agents in the network is known. The normal agents are equipped with a simple algorithm where neighbors taking the largest and smallest values are ignored to avoid being influenced by malicious agents. We address this problem when the inter-agent communication have bounded time-varying delays with an emphasis on networks of vehicles. Topological conditions in terms of graph robustness are developed.