基于信任度评估的分布式算力交易方案  

Distributed computing power trading scheme based on trustworthiness assessment

作　　者：王英晓春 王连海 徐淑奖 张淑慧 刘天瑞 WANG Yingxiaochun;WANG Lianhai;XU Shujiang;ZHANG Shuhui;LIU Tianrui(Key Laboratory of Computing Power Network and Information Security,Ministry of Education,Shandong Computer Science Center(National Supercomputer Center in Jinan),Qilu University of Technology(Shandong Academy of Sciences),Jinan 250014,China;Shandong Provincial Key Laboratory of Industrial Network and Information System Security,Shandong Fundamental Research Center for Computer Science,Jinan 250014,China)

机构地区：[1]齐鲁工业大学(山东省科学院),山东省计算中心(国家超级计算济南中心),算力互联网与信息安全教育部重点实验室,山东济南250014 [2]山东省工业网络和信息系统安全重点实验室,山东省基础科学研究中心(计算机科学),山东济南250014

出　　处：《网络与信息安全学报》2025年第1期79-91,共13页Chinese Journal of Network and Information Security

基　　金：国家自然科学基金(62102209);国家重点研发计划(2023YFC3304903);济南市“新高校20条”项目(202228017);泰山学者工程(tsqn202312231);山东省自然科学基金(ZR2024MF104,ZR2023QF129);齐鲁工业大学(山东省科学院)科教产融合试点工程重大创新类项目(2024ZDZX08);齐鲁工业大学(山东省科学院)人才项目(2023RCKY144)。

摘　　要：算力网络主要通过连接和整合分布式的异构计算节点,实现网络和计算资源的优化及高效利用。在算力网络中,节点因自身资源受限,通常需要以交易形式请求其他节点的计算资源来完成计算任务,引入信任度评估机制可以有效解决泛在异构节点所引起的节点可信度不高问题。然而,现有的算力交易方案虽然将信誉作为选择交易方的重要依据,但通常会选择信誉最高的节点,导致新加入节点在算力交易中无法享有公平参与交易的机会,且由于交易环境不可信,节点很难评估交易反馈的可信度。为了解决上述问题,提出了基于信任度评估的分布式算力交易方案,通过智能合约对满足算力需求的计算节点实施自动化的信任度评估,生成信任度降序列表,基于随机选择算法从信任度高于一定阈值的节点中选定交易方,以保证资源选择的随机性,有效防止合谋攻击。同时,利用适应性函数动态调整直接信任和间接信任的权重,以跟踪计算节点的行为变化。此外,借助区块链技术,设计了一种衡量反馈可靠性的方法,以减少因不可信节点提交恶意评分所造成的评估结果偏差。最后,通过实验证明了所提方案在识别恶意评分和降低恶意行为影响方面的有效性。The computing power network was primarily designed to optimize and efficiently utilize network and computing resources by connecting and integrating distributed heterogeneous computing nodes.Within this net‐work,nodes frequently required computing resources from other nodes through trading mechanisms due to their limited local resources.The introduction of a trust evaluation mechanism effectively addressed the issue of low trustworthiness among ubiquitous heterogeneous nodes.However,existing computing power trading schemes,which relied on reputation as a key criterion for selecting trading partners,typically favored nodes with the highest reputation.This approach resulted in newly joined nodes being unable to participate fairly in computing power trad‐ing,and the untrustworthy trading environment made it difficult for nodes to assess the credibility of trading feed‐back.To address these challenges,a distributed computing power trading scheme based on trust evaluation was pro‐posed.In this scheme,the trust evaluation of computing nodes meeting the computing power requirements was au‐tomated through smart contracts,generating a trust list sorted in descending order.A random selection algorithm was utilized to choose trading partners from nodes with trust levels above a predefined threshold,ensuring random‐ness in resource selection and effectively mitigating collusion attacks.Additionally,an adaptive function was em‐ployed to dynamically adjust the weights of direct and indirect trust,enabling the tracking of behavioral changes in computing nodes.Furthermore,a method for measuring feedback reliability was designed using blockchain technol‐ogy to reduce the bias in evaluation results caused by untrustworthy nodes submitting malicious ratings.Finally,ex‐perimental results demonstrated that the proposed scheme effectively identified malicious ratings and reduced the impact of malicious behaviors.The scheme is shown to ensure fair participation in trading for newly joined nodes while enhancing the rel

关 键 词：算力网络 区块链 信任度评估 算力交易 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]