Consensus over Clustered Networks with Asynchronous Inter-Cluster Communication

Within this work, the consensus problem for a clustered multi-agent system (C-MAS) is investigated. Given a multi-agent system (MAS), the agents are organized into disjoint clusters, where each cluster forms a connected network. The agents that are within the same cluster can communicate continuously with their neighbors. Between some cluster pairs, there exists an inter-cluster that enables the relay of information between the two clusters. Agents that have neighbors in clusters different from their own can communicate intermittently and asynchronously with their different-cluster neighbors. The goal of each agent in the C-MAS is to converge to an agreement value by sharing local information of a continuous-time homogeneous process. Note that the intermittent communication events of the continuous-time process are inherently hybrid. Therefore, a unique coupling between a static consensus protocol and a hybrid consensus protocol is designed. The closed-loop network model is presented using a hybrid systems framework. The consensus problem is then recast into a set stability problem and sufficient conditions of the consensus set are presented through leveraging a Lyapunov-based stability analysis. A C-MAS composed of 16 agents that is partitioned into 3 clusters and 2 inter-clusters is used to illustrate the results through numerical simulations and shown to achieve consensus using the proposed control strategy.