云计算环境下信任模型和框架研究

发布时间：2018-06-03 13:59

本文选题：云计算 + 信任模型　；参考：《合肥工业大学》2014年博士论文

【摘要】：云计算作为一种新兴的信息服务模式以及大规模数据存储和处理方式,正在为互联网时代服务计算带来巨大而深刻的变革,使得海量计算资源、存储资源、软件资源等通过互联网平台向外按需定制化提供,用户使用各种网络服务也变得更加方便而高效。然而,由于本身云计算环境具有巨大的开放性和复杂性,同时具有资源动态变化、自治性强、注重安全性等特征,用户在选择使用层出不穷的云服务时面临着各种安全、隐私等风险,同时受到各大云计算发起者爆出的各种安全事故的影响,逐渐引发了用户对云计算的信任危机,也阻碍了云计算的进一步普及和发展。在云计算面临各种信任问题的背景下,本文的研究目标是建立云计算信任管理机制,构建能够适应云计算环境与特点的信任模型,在开放和动态的云计算环境下对云服务信任度进行有效的管理和评估,从而降低用户选择云服务的风险。本文具体的研究内容包括： (1)提出了一种双层双视角的云服务信任评估模型——基于客观信任和主观信任的云服务信任评估模型,并且同时从局部和全局角度分别对云计算服务的信任度进行综合动态评估。在分布式信任服务提供商(TSP)的信任管理框架的基础上,从云服务的服务水平协议(SLA)记录信息和用户反馈信息两种角度,对云服务的局部主观信任(LST)、局部客观信任(LOT)、全局主观信任(GST)以及全局客观信任(GOT)进行评估,其中LST和LOT分别反映了从某云服务用户的单一视角对云服务提供商的主观信任及客观信任度,GST和GOT则反映了从全体用户视角对云服务提供商的主观信任及客观信任度。此外,对于多云环境,TSP之间需共享多云服务提供商在不同云中的信任信息,通过构建TSP之间的信任传播网络,从而能够对跨云环境下的云服务提供商可信度进行评估。通过仿真实验,表明我们提出的信任管理框架和评估方法在识别可信和不可信云服务提供商上是有效且健壮的。 (2)提出了一种云计算信任评估的三层信任属性框架,分别从软(硬)件等基础设施信任、平台及服务提供商的管理和技术服务信任以及用户服务交互提供信任上分析、提炼影响云服务信任度的各个信任属性。并且在此框架的基础上,提出了一种基于不同实体及不同视角差距模型的云服务可信评估方法。从具有相关专业经验的专家视角(即可信第三方)和具有个性化信任需求的用户视角考虑,分别提炼出云服务信任属性的实际提供性能(Delivery Performance)、用户感受性能(Perception Performance)以及重要性(Importance Performance)三者之间的差距,再基于差距模型从以上不同视角对云服务信任度进行了综合考量,得到云服务信任度的差距评估结果。该方法本质上是一种新颖的多属性决策模型,能够用来衡量用户与云服务提供者、可信第三方与云服务提供者以及用户与可信第三方之间对相同云服务的信任评估差异,从而找出影响云服务可信度的关键且薄弱属性,使得云服务提供商能够更加高效的提升服务质量,提高可信度。 (3)提出了一种新颖且有效的基于多属性信任评估的个性化云服务选择机制。云服务信任值的评估来自两个重要方面：基于感知的信任和基于信誉的信任。对于云服务用户来说，，需在使用服务之后向云服务系统反馈其服务交互的多属性评估结果,并存储在信任值数据库和信誉值数据库中,分别作为直接信任证据以供未来再次使用或提供给其他用户作为间接信任证据。在抽取了以上多属性的直接或间接信任证据之后,将基于感知的信任值和基于信誉的信任值聚合之后得到云服务信任值的最终结果。其中,基于信誉的信任值是在服务信誉值的基础上叠加一个从信誉值到信任值的映射函数而获取,该函数反应出用户对服务信誉的个性化偏好、偏见、信念等。通过这样一种基于个性化信任评估的云服务选择机制,本文提出的方法在云服务系统中能够有效地为用户选择符合其个性化信任需求的云服务。 (4)提出了一种基于反馈评价过滤机制的信任感知云服务推荐方法,首先将云服务的属性特征和用户需求模型中的各项需求偏好进行匹配,产生备选云服务。在信任构建平台所提供的信任反馈机制的基础上,通过结合反馈评分一致性和用户服务熟悉度两方面因素,对云服务不公平信任反馈评分进行了过滤。首先,反馈评分一致性原则是从反馈评分的内部规律出发,过滤掉偏离全体信任反馈评分平均水平较大的一部分评分；而用户服务熟悉度则是根据反馈评价者对于云服务的使用及反馈行为等外部规律,结合用户交互频率、服务使用时间、反馈提交时间等参数过滤。最终,结合内外两方面因素综合过滤不公平的反馈评价。
[Abstract]:As a new information service mode and a large data storage and processing mode, cloud computing is bringing huge and profound changes to the service computing in the Internet age. It makes the mass computing resources, storage resources, software resources and so on through the Internet platform to provide, and users use various network services as well. It is more convenient and efficient. However, because the cloud computing environment has huge openness and complexity, and has the characteristics of dynamic change of resources, strong autonomy and security, users face various security, privacy and other risks when choosing an endless cloud service. At the same time, they are exposed to various big cloud computing originators. In the background of cloud computing facing various trust problems, the aim of this paper is to establish cloud computing trust management mechanism, build trust model that can adapt to the environment and characteristics of cloud computing, and open and open. In the dynamic cloud computing environment, the trust of cloud services is managed and evaluated effectively, thus reducing the risk of user selection of cloud services.
(1) a double double perspective cloud service trust evaluation model, a cloud service trust evaluation model based on objective trust and subjective trust, is proposed, and the trust degree of cloud computing service is evaluated comprehensively and dynamically from both local and global perspectives. The foundation of trust management framework for distributed trust service provider (TSP) is based on the model of trust evaluation of cloud services. From two angles of service level protocol (SLA) record information and user feedback information, the local subjective trust (LST), local objective trust (LOT), global subjective trust (GST) and global objective trust (GOT) are evaluated, in which LST and LOT reflect the cloud service from a single perspective of a cloud service user. The subjective trust and objective trust of the provider, GST and GOT reflect the subjective and objective trust of the cloud service provider from the perspective of all users. In addition, for the multi cloud environment, the TSP needs to share the trust information of the multi cloud service provider in different clouds, and by building the trust communication network between the TSP, thus it can be able to cross the cloud. The credibility of the cloud service provider in the environment is evaluated. Through simulation experiments, it shows that our proposed trust management framework and evaluation methods are effective and robust in identifying trusted and untrusted cloud service providers.
(2) a three layer trust attribute framework for cloud computing trust evaluation is proposed. It provides trust analysis from soft (hard) components such as infrastructure trust, platform and service provider management, technical service trust and user service interaction, and extracts trust attributes that affect the trust degree of cloud services. A cloud service trustworthiness evaluation method based on different entity and different perspective gap model is proposed. The actual performance (Delivery Performance) of cloud service trust property (Delivery) and user perception performance (P) are extracted from the expert perspective of the relevant professional experience (that is, the letter of the letter) and the user's perspective of the personalized trust requirement. Erception Performance) and the gap between the importance (Importance Performance) three, and then based on the gap model to evaluate the cloud service trust from the above different perspectives, and get the result of the gap assessment of the cloud service trust. This method is essentially a new multi attribute decision model, which can be used to measure the user and the user. Cloud service providers, trusted third parties and cloud service providers, and trust assessment differences between users and trusted third parties to the same cloud services, thus identify the key and weak attributes that affect cloud service credibility, making the cloud service provider more efficient in improving the quality of service and improving credibility.
(3) a novel and effective personalized cloud service selection mechanism based on multi attribute trust evaluation is proposed. The evaluation of trust value of cloud services comes from two important aspects: perceived trust and reputation based trust. For cloud service users, the Xiang Yun service system needs to feed back the multiple genera of its service interaction after the service is used. The results of the assessment are stored in the trust value database and the reputation database as direct trust evidence for the future use again or provided to other users as indirect trust evidence. After extracting the direct or indirect trust evidence of the above multiple attributes, the perceived trust value and credit based trust value are aggregated. Then we get the final result of the trust value of the cloud service. Among them, the reputation based trust value is obtained on the basis of the service reputation value, which is superposed by a mapping function from the reputation value to the trust value. This function reflects the user's personalized preference, prejudice, belief and so on. The service selection mechanism proposed in this paper can effectively select cloud services that meet their personalized trust requirements in cloud service system.
(4) a trust aware cloud service recommendation method based on feedback evaluation and filtering mechanism is proposed. First, it matches the attribute features of the cloud service and the requirements preference of the user requirement model to generate alternative cloud services. On the basis of the trust feedback machine system provided by the trust construction platform, the consistency and consistency of the feedback score are combined with the feedback. In the two aspects of customer service familiarity, we filter the unfair trust feedback score of the cloud service. First, the principle of feedback score consistency is based on the internal rules of the feedback score, filtering out a part of the score that deviates from the average level of the whole trust feedback score, and the familiarity of the user service is based on the feedback evaluator. The use of cloud services and feedback behavior, such as the external rules, combined with user interaction frequency, service time, feedback time and other parameters filtering. Finally, combined with two aspects of internal and external factors to filter unfair feedback evaluation.
【学位授予单位】：合肥工业大学
【学位级别】：博士
【学位授予年份】：2014
【分类号】：TP393.08

【参考文献】