Hybrid-order Network Consensus for Distributed Multi-agent Systems

As an important field of Distributed artificial intelligence (DAI), multi-agent systems (MASs) have attracted the attention extensive research scholars. Consensus as most issue in MAS, much progress has been made studying consensus control but there are some problems remained largely unaddressed which cause MAS to lose useful network structure information. First, protocol usually proceeds over low-order by only considering direct edges between agents, ignores higher-order whole topology network. Second, existing work assumes all a same weight without exploring potential diversity connections. In this way, fail enforce consensus, resulting fragmentation into multiple clusters. To address above issues, paper proposes Motif-aware Weighted Multi-agent System (MWMS) method for control. We focus more on triangle motif network, it can be extended other kinds motifs well. novel weighted is used combination edge-based lower-order and motif-based structure, i.e., hybrid-order structure. Subsequently, simultaneously quantity quality connections framework designed update agents. Then, two baseline algorithms MWMS. our experiments, we use ten topologies different shapes, densities ranges comprehensively analyze performance proposed algorithms. The simulation results show that hybrid effectively enhance system topologies.