【Author】
Xie, Guangqiang; Xu, Haoran; Li, Yang; Hu, Xianbiao; Wang, Chang-Dong
【Source】INFORMATION SCIENCES
【Abstract】With the rapid development of wireless communication and localization technologies, multi-agent systems (MASs) have emerged as a powerful distributed artificial intelligence for consensus control. However, the connectivity of an MAS with a limited sensing range is vulnerable to the evolution of agents with high mobility, so converging to one common equilibrium rapidly while maintaining connectivity under a large-scale and high-density topology has been a research challenge. To address this problem, we develop a heuristic combinatorial algorithm that combines a distributed sector-division-based (SDB) consensus algorithm and a d-subgraph (DSG) connectivity maintenance algorithm. First, the communication region is heuristically divided into multiple sectors, and agents select representative neighbors to calculate control inputs by simultaneously considering the number and the distribution of perceived neighbors. Second, the d-subgraph is designed to update agents in accordance with constraint set constructed from the nearest neighbors in their locally perceived connected components. Thus, sequentially combining the SDB and the DSG algorithms affords a heuristic combinatorial (SDB&DSG) algorithm that can effectively accelerate convergence to a common equilibrium and retain global connectivity. The theoretical proofs of connectivity and convergence are given geometrically. Extensive simulations demonstrate the superiority of our algorithm, especially in large-scale and high-density topologies. (C) 2022 Elsevier Inc. All rights reserved.
【Keywords】Multi-agent system; Consensus seeking; Distributed control; Constraint set; Connectivity maintenance
【标题】在具有连通性维护的大规模和高密度多代理系统中快速寻求分布式共识
【摘要】随着无线通信和本地化技术的快速发展，多智能体系统（MAS）已经成为一种强大的分布式人工智能，用于共识控制。然而，具有有限传感范围的 MAS 的连通性容易受到具有高移动性的智能体进化的影响，因此在大规模和高密度拓扑下保持连通性的同时快速收敛到一个共同的平衡一直是一项研究挑战。为了解决这个问题，我们开发了一种启发式组合算法，它结合了基于分布式扇区划分 (SDB) 的共识算法和 d-子图 (DSG) 连通性维护算法。首先，通信区域被启发式地划分为多个扇区，代理通过同时考虑感知邻居的数量和分布来选择具有代表性的邻居来计算控制输入。其次，d-subgraph 被设计为根据从其本地感知的连接组件中的最近邻居构造的约束集来更新代理。因此，顺序组合 SDB 和 DSG 算法提供了一种启发式组合 (SDB&DSG) 算法，该算法可以有效地加速收敛到一个共同的平衡并保持全局连通性。几何上给出了连通性和收敛性的理论证明。广泛的模拟证明了我们算法的优越性，特别是在大规模和高密度拓扑中。 (C) 2022 Elsevier Inc. 保留所有权利。
【关键词】多代理系统；寻求共识；分布式控制；约束集；连接维护
评论