Research On SaaS Selection And Evolution Method Based On Trustworthiness Measurement

SaaS is the existence form of software service in cloud computing environment,and evolution is still one of its essential attributes.As more and more users choose SaaS as their information technology means,even if a single user occasionally makes an evolution request,in the face of many users,SaaS platform will face a large number of evolution requests at any time,and as a result SaaS evolution becomes an important part of the daily work of service providing platform.At the same time,SaaS service is different from traditional software because of its multi tenancy,on-demand billing,configurable and scalable features.This poses new challenges to SaaS services in terms of evolutionary quality and evolutionary efficiencyIn terms of evolution quality,add,replace and delete operations are the main implementation methods of evolution.In the face of many SaaS with the same or similar functions in the Cloud Market,how to find the implementation evolution of SaaS services that can meet the user's trustworthiness preferences and improve the overall trustworthiness of the service system has become a key issue to determine whether the evolution can succeed.In terms of evolution efficiency,SaaS,which is mainly obtained by renting,requires the service providing platform to respond to users' evolution requests quickly.Therefore,how to further simplify the methods and steps of SaaS service evolution becomes another key problem faced by SaaS.This thesis suggests that SaaS selection based on trustworthiness measurement results is an essential way to find the trustworthiness preference and ensure the quality of software evolution.In the aspect of evolution efficiency,in order to improve the evolution efficiency of SaaS,evolution research should not only focus on the evolution implementation method itself,but also treat the service trustworthiness measurement process,selection and acquisition process and evolution implementation process as an organism,so as to avoid the bottleneck of evolution efficiency caused by the inefficiency of a certain process.In view of the above contents,the contents of this thesis are summarized as follows.(1)Research on the measurement method of SaaS service trustworthiness.SaaS trustworthiness measurement is the basis of SaaS selection,which includes two aspects: SaaS independent trustworthiness measurement and SaaS combination trustworthiness measurement.In terms of the independent trustworthiness measurement of SaaS services,a new method based on information gain is proposed to calculate the weight of trusted sub attributes.Then,based on the weight of sub attributes,we propose a method to measure the service independent trustworthiness of SaaS based on information entropy and Markov chain.SaaS with high independent trustworthiness does not necessarily lead to the improvement of overall trustworthiness,but SaaS with low independent trustworthiness will inevitably lead to the reduction of overall trustworthiness.Therefore,it is one of the necessary conditions to measure the independent trustworthiness and select the SaaS with high independent trustworthiness to improve the overall trustworthiness of the service system.In the aspect of combination trustworthiness measurement,a method of SaaS service combination trustworthiness measurement based on Markov chain theory and cosine similarity is proposed.The external performance of the trustworthiness of SaaS service portfolio can be understood as the ability of services to resist the impact of external risk factors.That is to say,the smaller the change of the trustworthiness state of SaaS service portfolio is when it is affected by risk factors,the stronger the risk resistance ability is,and the higher the trustworthiness of the service portfolio is.The measurement results of SaaS combined trustworthiness can calculate the influence degree of different SaaS individuals on the overall trustworthiness.In this way,it can help the system finally select SaaS that can improve the overall trustworthiness to implement evolution,as a result,the quality of evolution can be guaranteed.(2)Aiming at the research of SaaS selection method based on evolution and trustworthiness measurement,the concept of trusted agent is proposed to realize the selection process of SaaS service.One end of the trusted agent connects to the services and service blocks to be evolved,to obtain the relevant information of the SaaS to be evolved for trustworthiness computing.The other end connects to the external SaaS service sets to obtain the information of each selected SaaS service in the external SaaS service collection for trustworthiness measurement.The trusted agent encapsulates the independent trustworthiness,combined trustworthiness and service selection algorithm of SaaS service,and realizes the measurement and selection process of SaaS service.Trusted agent realizes the connection between the internal and external service sets of evolution,and links the implementation process of service evolution with the process of measurement and selection,which provides a strong support for improving the efficiency and automation of evolution.(3)Research on the implementation method of SaaS evolution.The evolution process of SaaS includes static evolution and dynamic evolution.In the aspect of static evolution,this thesis proposes a description method of SaaS service architecture based on algebra of communication process(ACP).The internal architecture of service system is the basis of implementing dynamic evolution,which is conducive to quickly determine the location of evolution and the services affected by evolution.In the aspect of dynamic evolution,an evolutionary implementation method based on Petri net is proposed.The dynamic evolution is further simplified as the addition and deletion of the internal correlation of SaaS service system.By introducing the evolution execution center,the trusted agent is connected with the service sets outside,which realizes the combination of the trustworthiness measurement process,the service selection process and the evolution implementation process,so as to complete the dynamic evolution efficiently.