基于RippleNet的多维度算力节点安全信任评估方法  

Multi-dimensional security trust evaluation method for computing nodes based on RippleNet

作　　者：周犇 ZHOU Ben(China Mobile Group Design Institute Co.,Ltd.,Beijing 100080,China)

机构地区：[1]中国移动通信集团设计院有限公司,北京100080

出　　处：《电信工程技术与标准化》2024年第S02期133-136,共4页Telecom Engineering Technics and Standardization

摘　　要：算力网络安全是算力网络建设健康度的关键衡量指标。构成算力网络的算力节点安全作为系统整体安全的源头,目前缺乏专业的安全信任衡量评估标准来感知预测算力网络安全态势。本文提出了一种结合RippleNet的算力节点安全信任评估方法,用于评估算力网络节点的安全信任系数,重点关注节点计算资源和数据安全。对节点访问日志、安全配置、通信记录、安全事件等多性维度关联安全因子的发掘,提出了一套综合性的多维安全信任评估指标体系,推断每个节点的安全风险程度,从而提供有效的安全防护调度策略支撑。Computing network security is a critical indicator of the overall health of computing network infrastructure.The security of computing nodes,which make up the network,is fundamental to the system's overall security.Currently,there is a lack of professional standards for measuring and evaluating security trust,which hinders accurate perception and prediction of network security.This paper proposes a method for evaluating computing node security trust,utilizing RippleNet to assess the security trust coefficient of network nodes,with an emphasis on node computing resources and data security.By examining various security factors such as node access logs,security configurations,communication records,and security events,a comprehensive multi-dimensional security trust evaluation system is developed.This system aims to assess the security risk levels of individual nodes and provide effective support for security protection scheduling strategies.

关 键 词：算力网络 节点安全 信任评估 RippleNet 

分 类 号：TN918[电子电信—通信与信息系统]