ResilM-IBN:一种自愈型多智能体网络管理框架
ResilM-IBN:multi-agent framework for self-healing network management
摘要
Abstract
To address the challenges of inaccurate intent parsing,uncontrollable execution chains,and the lack of self-healing mechanisms in intent-based networking under multi-service environments,ResilM-IBN,a self-healing multi-agent management framework for software-defined network,aiming to enhance network intelligence and robustness,was proposed.Through structured communication and multi-agent collaboration,the framework ensures semantic consistency and task traceability throughout the intent execution process.A serial closed-loop execution process following the"intent-verification-execution-re-verification"workflow was constructed to improve controllability and reliability during intent realization.Furthermore,a collaborative self-healing mechanism was designed,where cross-agent anomaly detection and automated recovery were triggered by a dual-condition logic combining user-intent constraints and runtime feedback.Ex-perimental results obtained from a Mininet-Ryu prototype demonstrate that ResilM-IBN can achieve stable and reliable in-tent execution and self-healing recovery in complex scenarios.The proposed framework exhibits strong feasibility,stabil-ity,and portability,providing a practical reference for the engineering implementation of intent-based networking.关键词
意图驱动网络/软件定义网络/大语言模型/多智能体Key words
intent-based networking/software-defined network/large language model/multi-agent分类
信息技术与安全科学引用本文复制引用
董黎刚,苟敬文,蒋献,张子天,诸葛斌..ResilM-IBN:一种自愈型多智能体网络管理框架[J].通信学报,2026,47(1):250-266,17.基金项目
国家自然科学基金资助项目(No.W2421086,No.61871468) （No.W2421086,No.61871468）
浙江省科技创新重点基金资助项目(No.2023R5211) （No.2023R5211）
浙江省自然科学基金资助项目(No.LZ23F010003) （No.LZ23F010003）
浙江省重点研发计划基金资助项目(No.2026C02A1244) （No.2026C02A1244）
桐乡通用人工智能研究院基金资助项目(No.TAGI2-B-2024-0014) （No.TAGI2-B-2024-0014）
浙江省新型网络标准及应用技术重点实验室基金资助项目(No.2013E10012)The National Natural Science Foundation of China(No.W2421086,No.61871468),Key Scientific and Techno-logical Innovation Project of Zhejiang Province(No.2023R5211),The Natural Science Foundation of Zhejiang Province(No.LZ23F010003),Key Research and Development Program of Zhejiang Province(No.2026C02A1244),Tongxiang General Artifi-cial Intelligence Research Institute Project(No.TAGI2-B-2024-0014),Key Laboratory of Emerging Network Standards and Applica-tion Technologies of Zhejiang Province(No.2013E10012) （No.2013E10012）
