Trust Model And Framework Research In Cloud Computing Environment

As an emerging information service paradigm, Cloud Computing is bringing great and deep revolution to Service Computing in the Internet era, through which massive computing resources, storage resources, software resources, etc., can be delivered as customized services on demand, and customers can use a variety of web services in a more convenient and efficient manner. However, due to the great openness and com-plexity in Cloud Computing environment, and also features of dynamics in the re-sources, strong autonomy, and high security consciousness, customers confront various security and privacy risks when selecting cloud services. In addition, affected by all kinds of security accidents in many Cloud Computing vendors, it triggers customers’ trust crisis to Cloud Computing, and thus obstacle the further spread and development of Cloud Computing. Under the background of a variety of trust problems in Cloud Computing environment, the research objective of this paper is to establish trust models adapted to the characteristics of Cloud Computing, build Cloud Computing trust man-agement mechanism, and evaluate and manage cloud services’trust effectively. We hope that the cloud service trust mechanism can be introduced by this research, by which the cloud service’s trustworthiness can be reliably evaluated and managed under such open and dynamic environment, and thus decrease the customers’risks of choosing cloud services. The detailed research contents can be concluded as follows:(1) We propose a bi-layer and bi-prospective cloud service trustworthiness evaluation model, which is based on objective trust and subjective trust model, to dynamically evaluate cloud service trustworthiness in the local and global layer respectively. On the basis of trust management framework of Trust Service Provider (TSP), the SLA record information and users’feedback of cloud services are collected, to evaluate the cloud services’local subjective trust (LST), local objective trust (LOT), global subjective trust (GST), and global objective trust (GOT). LST and LOT are the subjective trust and ob-jective trust of the cloud services from one particular cloud service user, respectively, and GST and GOT are the subjective trust and objective trust of the cloud services from all the cloud service users. In addition, under multi-cloud environment, TSPs can share multi-cloud service providers’trust information in different clouds and build trust prop-agation network among TSPs, so as to evaluate the trustworthiness of cloud service providers in multi-cloud environments. The simulation experiments show that our pro- posed trust management framework and evaluation method are effective and robust in differentiating trustworthy and untrustworthy service providers.(2) We propose a three-layer trust attribution framework for cloud trustworthiness evaluation, and analyze and refine attributions that have impact on cloud trustworthi-ness from the software (hardware) infrastructure trust layer, management and techno-logical trust layer of platforms and service providers, and user interaction trust layer. On the basis of this framework, we propose a cloud trustworthiness evaluation approach based on the gap model from different entities’different perspectives. Considering the perspective from experts who have professional experiences (which are deemed as trustworthy third party) and the perspective from users who have personalized QoS de-mand, the delivery performance and perception performance of cloud services are de-rived, both of which are compared to the importance performance of cloud service trustworthiness attribution. Thus, the trustworthiness of cloud services is evaluated through gap model comprehensively. This method is essentially a novel mul-ti-attribution decision making model, which can be used to measure the difference of the evaluation by users and service providers, so that the key but week attribution affecting the cloud service trustworthiness can be found, and help the cloud service provider en-hance the QoS and increase their trustworthiness more efficiently.(3) We propose a novel and effective cloud service selection mechanism based on multi-attribution cloud trustworthiness evaluation. The cloud service trustworthiness evaluation comes from two importance aspects:perception-based trust and reputa-tion-based trust. For the cloud service users, they are supposed to return their service interaction feedback to the cloud service system, and the feedback is stored in the trust value base and reputation value base, respectively, which are to be used in the future as direct trust evidence or to be used by other users as indirect trust evidence. After the multi-attribution direct or indirect trust evidence is elicited, the cloud service selection mechanism applies Evidential Reasoning approach to form the final trust evaluation results. The final cloud service trust value is the aggregate value of perception-based trust value and reputation-based trust value, and the reputation-based trust value is ac-quired by the mapping function from the reputation value to reputation-base trust value. The mapping function reflects the personalized preferences, bias, belief, etc., of users to services’reputation. Through such cloud selection mechanism based on personalized trustworthiness evaluation, our proposed method can select cloud services which are in accordance with the personalized trust requirements of users in cloud service systems. (4) We propose a trust-aware cloud service recommendation method based on feed-back rating filtering mechanism. Firstly, match the attribution characteristics of the cloud services and each requirement preference in users’requirement model to generate alternative cloud services. Then, based on the trust feedback rating mechanism provided by Trust Establishing platform, filter the unfair trust feedback ratings combining feed-back rating consistency and users’familiarity to the cloud services. The rule of feedback rating consistency is based on the endogeneous trend of the trust feedback ratings to filter the ratings which are deviated from most other ratings. The familiarity factor is based on the-exogenous rule of the feedback rators’ rating behaviofto filfer the untrust-worthy rator’s ratings considering the factors such as interaction frequency, service us-ing time, feedback submit time, etc. Ultimately, unfair feedback ratings are filtered combining these two endogeneous and exogenous factors.