Secure Consensus with Distributed Detection via Two-hop Communication

Abstract: In this paper, we consider a multi-agent resilient consensus problem, where some of the nodes may behave maliciously. The approach is to equip all nodes with a scheme to detect neighboring nodes when they behave in an abnormal fashion. To this end, the nodes exchange not only their own states but also information regarding their neighbor nodes. Such two-hop communication has long been studied in fault-tolerant algorithms in computer science. We propose two distributed schemes for detection of malicious nodes and resilient consensus with different requirements on resources for communication and the structures of the networks. In particular, the detection schemes become effective under certain connectivity properties in the network so that the non-malicious nodes can share enough information about their neighbors. It is shown that the requirements are however less stringent than those for conventional algorithms. A numerical example is presented to demonstrate the performance of the proposed methods in wireless sensor networks.