【作者】 陈雪；

【导师】 唐碧华；

【作者基本信息】 北京邮电大学 ， 电子与通信工程（专业学位）， 2022， 硕士

【摘要】 2016年,欧洲电信标准协会(ETSI)将边缘计算的概念从移动边缘计算(Mobile Edge Computing)拓展为多接入边缘计算(Multi-Access Edge Computing),并将边缘计算从电信蜂窝网络进一步延伸至其他无线接入网络,使得各种异构设备可加入网络并提供计算服务。虽然边缘计算网络的健壮性得到加强,但加入网络的边缘节点容易受到攻击,导致网络面临安全威胁。然而,无论是移动边缘计算还是多接入边缘计算,现有的研究中鲜少考虑终端用户从边缘服务节点所获得的计算结果的正确性,尤其是容易被攻击的多接入边缘计算,边缘节点的服务可靠性无法得到充分的保证。本文将区块链技术和零知识证明技术引入边缘计算网络,旨在不过度消耗网络计算资源的前提下,能够实现在不泄露计算结果的同时,验证边缘节点计算结果的正确性,从而提高边缘计算的安全性。其研究内容如下:基于区块链、智能合约和零知识证明等技术,本文提出了一种新型的安全边缘计算方案。该方案利用区块链去中心化信任的特点,实现对边缘服务节点的信任。但由于通过区块链节点冗余计算实现的服务可靠性保护,不完全适用于计算密集的边缘计算场景,本文创新性地在此方案中加入零知识证明技术。边缘服务节点将在链下处理任务请求,此过程无需向区块链网络广播,并在链上调用零知识证明智能验证合约对执行结果进行验证,以达到提高边缘节点服务可靠性的目的。以TSP(travelling salesman problem)为演示实例,本文在实物演示验证平台上进行了实际测试,并与直接通过区块链节点冗余计算实现边缘节点可靠服务的方案进行了对比分析。测试结果表明,对于单个区块链节点,同时使用区块链技术和零知识证明技术所消耗的gas,仅为直接使用区块链技术的1/150,本文提出的方案达成了在提高边缘计算安全性的同时,只带来相对较小的资源消耗的目标。为了满足边缘计算业务的各种时延要求,以及进一步实现安全边缘计算,本文设计了基于零知识证明的分级信誉值算法ZKP-Reputation。该算法基于零知识证明对任务处理结果的验证情况,综合考虑时间、犯错次数、计算量大小等多种因素,管理服务节点的信誉值。对于信誉值较高的边缘服务节点,终端用户可直接信任其计算结果,无需等待零知识证明验证合约调用结果的反馈,以减小加入区块链和零知识证明技术带来的时延的影响。对ZKP-Reputation算法的仿真测试结果表明,终端用户根据ZKP-Reputation算法结果选择信誉值较高的服务节点直接信任其处理结果时,累计进行20次计算所消耗的总时间,仅为不使用ZKP-Reputation算法时所用总时间的27.5%。因此,加入信誉值算法,有效降低了加入区块链和零知识证明带来的影响。根据提出的安全边缘计算方案和ZKP-Reputation算法,本文搭建了由15个节点组成的实物演示验证平台。此平台采用cosmos为系统提供底层区块链服务,利用truffle实现信誉值合约和零知识证明验证智能合约部署,并通过内网穿透将区块链节点互联组网,完成测试准备。平台搭建完成后对系统的功能以及零知识证明的资源消耗进行了测试,以确保系统的可用性。测试结果说明了将区块链技术和零知识证明技术应用于边缘计算网络可以保证边缘服务节点的计算结果可靠,同时还实现了对计算结果的隐私性保护。更多还原

【Abstract】 In 2016,the European Telecommunications Standards Institute(ETSI)expanded the concept of Edge Computing from Mobile Edge Computing to multi-access Edge Computing.Moreover,edge computing is extended from telecom cellular network to other wireless access network so that various heterogeneous devices can join the network and provide computing services.Although the robustness of edge computing network is enhanced,edge nodes joining the network are vulnerable to attack and thus face security threats.However,whether it is mobile edge computing or multiaccess edge computing,the correctness of the calculation results obtained by end-users from edge service nodes is not taken into account in the existing researches,especially in multi-access edge computing,which is easy to be attacked,and the reliability of edge node service cannot be fully guaranteed.In this thesis,block chain technology and zero-knowledge proof technology are introduced into edge computing network,aiming to verify the correctness of edge node calculation results without leakage of calculation results without excessive consumption of network computing resources,so as to improve the security of edge computing.The research contents are as follows:Based on blockchain,smart contract and zero-knowledge proof,a new security edge computing scheme is proposed in this paper.This scheme utilizes the decentralized trust characteristics of blockchain to realize the trust of edge service nodes.However,since the protection of service reliability achieved by redundant computing of blockchain nodes is not fully applicable to compute-intensive edge computing scenarios,this paper innovatively adds zero-knowledge proof technology into the scheme.The edge service node will process the task request under the chain without broadcasting to the blockchain network,and the zero knowledge proof intelligent verification contract will be invoked on the chain to verify the execution result,so as to improve the service reliability of edge node.Taking TSP(Travelling Salesman Problem)as a demonstration example,this paper carries out a practical test on the physical demonstration platform,and makes a comparative analysis with the scheme of realizing edge node reliable service directly through the redundancy computing of blockchain nodes.The test results show that for a single blockchain node,the gas consumed by using both blockchain technology and zero-knowledge proof technology is only 1/150 of that consumed by using blockchain technology directly.The scheme proposed in this paper achieves the goal of improving the security of edge computing while only bringing relatively small resource consumption.In order to meet various delay requirements of edge computing services and further realize secure edge computing,this paper designs a hierarchical Reputation algorithm based on zero-knowledge proof,ZKPREPUTATION.Based on the verification of task processing results with zero-knowledge proof,the algorithm comprehensively considers many factors,such as time,error times and computation amount,to manage the credit value of service nodes.For edge service nodes with high reputation,end users can directly trust their calculation results without waiting for the feedback of the call result of zero-knowledge proof to verify the contract,so as to reduce the impact of the delay caused by the addition of blockchain and zero-knowledge proof technology.The simulation test results of ZKPReputation algorithm show that when the end user chooses the service node with high Reputation value according to the results of ZKP-Reputation algorithm to directly trust its processing results,the total time consumed by 20 calculations is accumulated.Only 27.5%of the total time when zKPReputation algorithm is not used.Therefore,the addition of the credit algorithm effectively reduces the impact of the addition of blockchain and zero-knowledge proof.According to the proposed security edge calculation scheme and ZKP-Reputation algorithm,this paper builds a physical demonstration and verification platform composed of 15 nodes.Cosmos was used to provide the underlying blockchain service for the system,Truffle was used to implement the credit contract and zero-knowledge proof to verify the deployment of smart contract,and blockchain nodes were interconnected through Intranet penetration to complete test preparation.After the platform was built,the function of the system and the resource consumption of zero knowledge proof were tested to ensure the availability of the system.The test results show that the application of block chain technology and zero-knowledge proof technology to edge computing network can ensure the reliability of computing results of edge service nodes,and realize the privacy protection of computing results.更多还原