An Online Topology Control Method for Multi-Agent Systems: Distributed Topology Reconfiguration
Yu Feng, Fuyong Wang, Zhongxin Liu, Zengqiang Chen
Abstract
The distributed methodology for multi-agent systems to autonomously control their network topologies holds significant promise for handling complex and changeable tasks, and is collectively referred to as online topology control. In this article, we delve into the feasibility of online topology and introduce a novel approach called Distributed Topology Reconfiguration (DTR). Inspired by cellular automata, DTR aims to design a unified strategy that enables agents to adaptively regulate their local topology, thereby achieving online optimization of the overall topology to meet task requirements. To actualize this idea, we mathematically characterize the evolution of network topology considering both internal dynamics and external control inputs, and clarify the topology control objective by defining a new form of objective function. To steer the evolution of topology towards the desired pattern, we propose a DTR method, which involves designing a unified control strategy for each node to generate control inputs based on its observed local subgraphs. Additionally, we establish a sufficient condition for ensuring the stability and effectiveness of the topology system. Finally, we present several application scenarios of online topology control and develop corresponding DTR strategies to meet the respective requirements. The effectiveness of these strategies is demonstrated through theoretical analysis and numerical simulation.