摘要
多域软件定义无线网络(SDWN)中的跨域组播路由问题不仅是NP难组合优化问题,随着网络规模的增加和组播组成员的动态变化,构建高效的跨域组播路由路径还需要及时灵活获取和维护全局网络状态信息并设计出最优跨域组播树问题的求解算法。针对现有求解方法对网络流量状态感知性能欠缺影响组播业务对QoS方面需求的满足,并且收敛速度慢难以适应网络状态高度动态变化的问题,本文设计和实现了一种基于多智能体深度强化学习的SDWN跨域组播路由方法(MA-CDMR)。首先,设计了组播组管理模块和多控制器之间的通信机制来实现不同域之间网络状态信息的传递和同步,有效管理跨域组播组成员的加入和离开;其次,在通过理论分析和证明最优跨域组播树包括最优的域间组播树和域内组播树两个部分的结论后,本文对每个控制器设计了一个智能体,并设计了这些多智能体之间的协作机制,以保证为跨域组播路由决策提供网络状态信息表示的一致性和有效性;然后,设计一种在线与离线相结合的多智能体强化学习训练方式,以减少对实时环境的依赖并加快多智能体收敛速度;最后,通过系列实验及其结果表明所提方法在不同网络链路信息状态下具有达到了很好的网络性能,平均瓶颈带宽相较于现有KMB、SCTF、DRL-M4MR和MADRL-MR方法分别提升了7.09%、46.01%、9.61%和10.11%;平均时延在与MADRL-MR方法表现相近的同时,相比KMB、SCTF和DRL-M4MR方法有明显提升,而丢包率和组播树平均长度等也均优于这些现有方法。本文工作源代码已提交至开源平台https://github.com/GuetYe/MA-CDMR。
The cross-domain multicast routing problem in a software-defined wireless network(SDWN)with multi-controller domains is a classic NP-hard combinatorial optimization problem.As the network size increases,the design and implementation of cross-domain multicast routing paths in this network necessitates not only the development of efficient algorithms for identifying the optimal cross-domain multicast tree but also the assurance of timely and flexible acquisition and maintenance of global network state information.To address the shortcomings of existing solutions have limit ability to sense the network traffic state,affecting the quality of service(QoS)of multicast services,and the difficulty adapting to the highly dynamically changing network state and slow convergence speed,this paper designs and implements a multiagent deep reinforcement learning-based cross-domain multicast routing(MA-CDMR)method for the SDWN with multicontroller domains.First,a multi-controller communication mechanism and a multicast group management module are designed to transfer and synchronize network information between different control domains of the SDWN,respectively,thus effectively managing the joining and leaving of members in the cross-domain multicast group.Second,a theoretical analysis and proof show that the optimal cross-domain multicast tree includes an interdomain multicast tree and an intradomain multicast tree.An agent is set up for each controller,a cooperation mechanism between multiple agents is designed to effectively optimize cross-domain multicast routing and improve the overall network performance and resource utilization,and a multiagent reinforcement learning-based method that combines online and offline training is designed to reduce the dependence on the real-time environment and increase the convergence speed of multiple agents.Finally,a series of experiments and their results show that the proposed method achieves good network performance under different network link information states,and the average bottleneck bandwidth is improved by 7.09%,46.01%,9.61%,and 10.11%compared with the existing KMB,SCTF,DRL-M4MR,and MADRL-MR methods,respectively;the average delay is similar to that of MADRL-MR,and significantly better than that of KMB,SCTF,and DRLM4MR;the packet loss rate and the average length of multicast tree are also better than those of these existing methods.The average delay is similar to that of the MADRL-MR method while it is significantly improved over the KMB,SCTF and DRL-M4MR methods,while the packet loss rate and the average length of the multicast tree are also better than these existing methods.The source code of this work has been made available on the open-source platformhttps://github.com/GuetYe/MA-CDMR.
作者
叶苗
胡洪文
王勇
何倩
王晓丽
文鹏
郑基浩
YE Miao;HU Hong-Wen;WANG Yong;HE Qian;WANG Xiao-li;WEN Peng;ZHENG Ji-Hao(Key Laboratory of Cognitive Radio and Information Processing of Ministry of Education,Guilin University of Electronic Technology,Guilin,Guangxi 541004;School of Computer Science and Information Security,Guilin University of Electronic Technolgoy,Guilin,Guangxi 541004;School of Computer Science and Technology,Xidian University,Xi'an 710071)
出处
《计算机学报》
北大核心
2025年第6期1417-1442,共26页
Chinese Journal of Computers
基金
国家自然科学基金项目(Nos.地区62161006,面上62372353,重点U22A2098)
认知无线电与信息处理教育部重点实验室主任基金(No.CRKL220103)
广西研究生教育创新计划项目(No.YCBZ2023134)资助。
关键词
组播树
软件定义无线网络
跨域组播路由
多智能体
深度强化学习
multicast tree
software defined wireless network
cross domain multicast routing
multi agent
deep reinforcement learning
