Event triggering strategies for consensus in clustered networks

This paper focuses on consensus in networks partitioned in several clusters. It uses the multi-agent framework in which the network is seen as a sum of interconnected subsystems called agents. We assume that each agent updates its state continuously by taking into account the states of some other agents belonging to the same cluster. This protocol allows reaching only local agreements in the network. In order to get consensus we endow an agent per cluster with the capacity to discretely interact outside its own cluster. The discrete interaction of one agent with agents from other clusters is modeled as a state jump or reset. The inter-clusters interactions are activated by using event dependent rules. The goal of the paper is to design event triggering reset strategies that guarantee the consensus is achieved. No global dwell time separating the reset instants is imposed but we show that a dwell time per cluster is ensured by the proposed reset strategies.