Consensus Seeking in Large-Scale Multiagent Systems With Hierarchical Switching-Backbone Topology

Guangqiang Xie,Haoran Xu,Yang Li,Chang-Dong Wang,Biwei Zhong,Xianbiao Hu

DOI: https://doi.org/10.1109/tnnls.2023.3290015

IF: 14.255

2023-01-01

IEEE Transactions on Neural Networks and Learning Systems

Abstract:Recent developments in multiagent consensus problems have heightened the role of network topology when the agent number increases largely. The existing works assume that the convergence evolution typically proceeds over a peer-to-peer architecture where agents are treated equally and communicate directly with perceived one-hop neighbors, thus resulting in slower convergence speed. In this article, we first extract the backbone network topology to provide a hierarchical organization over the original multiagent system (MAS). Second, we introduce a geometric convergence method based on the constraint set (CS) under periodically extracted switching-backbone topologies. Finally, we derive a fully decentralized framework named hierarchical switching-backbone MAS (HSBMAS) that is designed to conduct agents converge to a common stable equilibrium. Provable connectivity and convergence guarantees of the framework are provided when the initial topology is connected. Extensive simulation results on different-type and varying-density topologies have shown the superiority of the proposed framework.

computer science, artificial intelligence, theory & methods,engineering, electrical & electronic, hardware & architecture