Interaction Reduction and Consensus for Multi-agent Systems

This paper studies a coordination strategy to utilize less data received from interactions among agents in a network. A distributed scheme is presented for each agent to perform interaction reduction independently using only the status information from its neighbors. As a result of the proposed scheme, agents can make their own decisions locally for a coordination purpose by ignoring the data received through deleted links. A consensus coordination problem is used as an example in the paper to illustrate the effect of interaction reduction on steady states and transient response of the consensus process.