Weighted Consensus Protocols Design Based on Network Centrality for Multi-Agent Systems With Sampled-Data

Abstract

This technical note aims at constructing and analyzing an efficient framework for the leader-following consensus protocol in multi-agent systems (MASs). We propose two novel consensus protocols weighted by calculating the betweenness and eigenvector centralities for agent and link which are determined by the interconnection structure of MASs. The concepts of centrality were introduced in the field of social science. Ultimately, the use of the proposed protocols can be described with regard to not only the number of each agent's neighbors, which was utilized in the existing works, but also more information about agents through considering two such centralities. By utilizing the Lyapunov method and some mathematical techniques, the leader-following guaranteed cost consensus conditions for MASs with the proposed protocols and sampled-data will be established in terms of linear matrix inequalities (LMIs). Based on the result of consensus criteria, two new protocol design methods which utilize the betweenness and eigenvector centralities will be proposed. Finally, some simulation results are given to illustrate the advantages of the proposed protocols in point of the robustness on sampling interval and the transient consensus performance.