【作者】 谢辉；

【导师】 郑宏；

【作者基本信息】 电子科技大学 ， 控制科学与工程， 2021， 硕士

【摘要】 新时代背景下物联网和5G通信技术正处于急速发展的状态,位于网络边缘的终端设备数量规模越来越庞大,海量数据的计算和处理给传统的云计算带来了极大的压力。边缘计算是在离终端设备距离更近的地方来提供相应的服务,使得云计算面临的数据传输延时和带宽等压力得到缓解。但是,边缘计算中的网络设备也面临着各个方面的安全挑战。云计算模式下的常规防护机制,难以有效保护计算和存储资源都较为有限的终端设备,而且终端设备数量规模庞大且分布广泛,一旦被感染的终端设备接入到边缘计算平台,会导致各种安全问题。因此,如何有效防护边缘计算环境中的接入安全非常关键。针对上述问题,本文针对边缘计算环境下的可信接入安全技术进行研究,设计并实现了一种基于身份的匿名认证密钥协商(AAKA)协议和一种基于BP神经网络的信任度评估模型,具体研究内容如下:1.通过分析边缘计算环境中终端设备和边缘计算服务器相互认证所面临的安全挑战,研究并设计了一种适用于边缘计算环境的基于身份的匿名认证密钥协商协议。该协议基于椭圆曲线加密系统和双线性配对等理论知识,终端设备和边缘计算服务器仅需要在云端注册中心经过注册后即可进行相互认证,认证过程中只需一个来回的通信流程。基于复杂性假设和随机预言机模型的安全性证明表明,该协议自身具备较高的安全性,并且具备终端设备匿名性、前向保密性等安全特性。2.通过研究边缘计算环境中的信任度机制,选择合适的信任度评估因子,基于BP神经网络模型构建了一个信任度评估模型。首先通过Edge Cloud Sim仿真软件获取数据来进行仿真实验,再通过对比其他三种分类方法来验证BP神经网络模型的评估性能优越性,仿真结果说明我们的信任度评估模型在评估性能方面取得了更好的表现。3.最后基于边缘计算开源平台Baetyl实现了可信接入安全模块,在测试环境中验证了本文所设计的身份认证协议的安全性,并通过对比表明了该协议在传输成本和计算成本上都取得了更佳的性能优势。针对信任度评估模块,对比AHP层次分析法进行实验,结果表明本文中基于BP神经网络的信任度评估方法取得了更佳的评估准确率。更多还原

【Abstract】 In the context of the new era,the Internet of Things and 5G communication technologies are in a state of rapid development.The number of terminal devices at the edge of the network is getting larger and larger.The calculation and processing of massive data has brought great pressure to traditional cloud computing.Edge computing is to provide corresponding services in a place closer to the terminal equipment,so that the pressure of data transmission delay and bandwidth faced by cloud computing is alleviated.However,network device in edge computing also faces various security challenges.The conventional protection mechanism in the cloud computing model is difficult to effectively protect terminal devices with limited computing and storage resources,and the terminal devices are large in scale and widely distributed.Once the infected terminal device is connected to the edge computing platform,it will lead to various kinds of security problem.Therefore,how to effectively protect access security in the edge computing environment is very critical.In view of the above problems,this paper studies the trusted access security technology in the edge computing environment,and designs and implements an identity-based anonymous authentication key agreement(AAKA)protocol and a BP neural network-based trust evaluation model,The specific research content is as follows:1.By analyzing the security challenges in the process of mutual authentication between terminal devices and edge computing servers in the edge computing environment,research and design an identity-based anonymous authentication key agreement protocol suitable for edge computing environments.The protocol is based on theoretical knowledge such as the elliptic curve encryption system and bilinear pairing.The terminal device and the edge computing server only need to be registered in the cloud registry to perform mutual authentication,and only a round-trip communication process is required during the authentication process.The security proof based on the complexity assumption and the random oracle model shows that the protocol itself has high security,and has security features such as terminal device anonymity and forward secrecy.2.By studying the trust mechanism in the edge computing environment and selecting appropriate trust evaluation factors,a trust evaluation model is constructed based on the BP neural network model.First,we use Edge Cloud Sim simulation software to obtain data for simulation experiments,and then compare the other three classification methods to verify the evaluation performance superiority of the BP neural network model.Finally,the simulation results show that our trust value evaluation model has achieved better evaluation performance.3.Finally,a trusted access security module is implemented based on the edge computing open source platform Baetyl.The security of the identity authentication protocol designed in this paper is verified in the test environment,and the comparison shows that the protocol has achieved a better performance in both transmission and calculation costs.For the trust evaluation module,the experiment is compared with the AHP analytic hierarchy process.The results show that the trust evaluation method based on the BP neural network in this paper has achieved better evaluation accuracy.更多还原