| Traditional computing environment is so static and closed that it is difficult to cope with thenew challenges of complex network environment. In this case, service computing comes, inservice computing environment, the system can adapt to environmental changes and businessneeds dynamically. Service computing makes web services as the basic structure unit. Webservices have open, cross-platform, and highly dynamic characteristics, which makes it veryvulnerable to security attacks, and thus the space for development of Web services andapplication depth will depend largely on its own the level of safety, especially the ability toprovide an effective security mechanism to prevent the malicious visit or call by non-authorizedusers.Although traditional access control mechanisms can make access decisions according tosecurity policy to some extent, the network environment has become increasingly open andcomplex, user groups are no longer fixed, the system cannot establish access control listseparately for each user. Therefore, the article makes security issues in computing environmentas a center, research trust models and access control model, the main work and results are asfollows:First, analyze the development of access control and key technologiesThis paper introduced security threat in services computing environment and importance ofresearching services security today, described the development of access control model, trustassessment mechanisms and trust model research, compared the advantages, disadvantages ofvarious models, the strengths and weaknesses of each trust evaluation mechanism. Relatedtheoretical study based on services computing and its features, the Web service systemarchitecture and so on.Second, embed SAML authentication module and XACML access control module in themodelService providers cannot fully use trust information from the service requesters for safetyassessment, authorization and access control based on trust evaluation results, the article putsforward a new kind of trust-based access control model. The model combines the SecurityAssertion Markup Language (SAML) with extensible Access Control Markup Language(XACML), embeds SAML authentication module and XACML access control module in themodel, integrates the authentication mechanism into the traditional trust model, makes accesscontrol more flexible reliable.Third, set up trust-based service matchmaking access control modelThe network load is the serious problems at this stage, network resources are not fully utilized,80%users use only20%network services, based on the network status, the level of trust isdivided into different levels, services provided to the service requester are able to meet theirneeds and the lowest level of trust services, so that you can set aside a higher level of service todemand a higher level of trust service requester. Fourth, predict trust by exponential smoothing algorithmUsing the exponential smoothing algorithm for prediction of trust, using historical sequencegenerated by credibility to predict the future the value of trust, so that select the user havingappropriate trust value to interact.Fifth, propose three-dimensional trust evaluation mechanismFor today based on the two-dimensional trust evaluation mechanism (direct view,recommend view), this paper proposed three-dimensional trust evaluation mechanism (directview, recommended view and public view), public view is the kind of public opinion, select themost recent period of good performance the three individuals as a monitoring group, the trust ofthe service requester to tell service providers, three members of the organization regularlyupdated in order to adapt to the dynamic changes in the network environment.Sixth, cluster algorithm to search the closest recommendersEvaluation of the same service from different users may be different because of its differentservice concerns, therefore, when obtaining the recommended information, should take serviceconcerns of users into consideration, the trust evaluation of the same service with focus closeenough is more valuable. Chapter5provided a cluster algorithm to search the recommenders,three-dimensional information of individual users (name, service, attention) is stored in thedatabase, provide recommendations according to the services used by individual users and theirthe concerns of the service, select the service requester and the same concerns or closest to theuser as recommended, recommended information provided by this user is the most accurate.Seventh, design simulation experiments to verify the effectiveness of experimentDesign simulation experiments to verify the accuracy and operational efficiency of theproposed algorithm, the evaluation criteria of the accuracy is square error (using the proposedalgorithm on a score with the actual score variance), the evaluation standard of recommendationalgorithm efficiency is trust computational overhead (an interaction with the average time tocomplete). The design of an actual case, the ideas presented in this article is reflected in theactual system, to some extent, is the combination of theory and practice. |