Consensus over Clustered Networks Using Output Feedback and Asynchronous Inter-Cluster Communication

This paper develops a method to yield state consensus and state reconstruction for a clustered multi-agent system (C-MAS). The agents within the network are organized into disjoint clusters, where each cluster induces a connected sub-graph. Agents contained within the same cluster are capable of communicating continuously with their neighbors. Between some cluster pairs, there exists an inter-cluster, where agents contained within the same inter-cluster can intermittently communicate with their neighbors. In addition, the communication between distinct inter-clusters may be asynchronous. Since we assume no agent can completely measure its own state, a model-based observer utilizing intermittent output feedback is developed to facilitate state reconstruction. The combination of continuous-time dynamics with intermittent communication and sensing is modeled as a hybrid system. The state consensus and state reconstruction problems are formulated as set stabilization problems for hybrid dynamical systems, where Lyapunov-based stability analyses show that the state consensus and state reconstruction sets are globally exponentially stable.