A Method Towards QoS-aware Web Service Composition Algorithm In Trusted Service Architecture

The Internet is characterized by complexity,anonymity and dynamics.Therefore,the problem of Web Service Composition not only needs to consider service consumer's requirements and global constraints,but also needs to consider the dynamic nature of Qo S(Quality of Service)attributes and the credibility of service composition environment.Based on Blockchain technology and fuzzy mathematical method,this thesis studies how to construct of trusted Web service composition environment to support flexible composition of Web services,which has theoretical research significance and strong applied value.The main work and innovative contributions of this thesis are as follows.(1)Analyzing the problems faced by trusted service composition and studying the framework of Web service composition.Based on the lifecycle of Web service composition,a trusted service composition framework is realized using Blockchain technology.This framework mainly consists of five modules: Basic Layer,Constraint Layer,Combination Layer,Verification and Monitoring Layer,and Smart Contract Layer.Each module is explained and analyzed in detail through diagrams.(2)There are multiple types of representation method of Web service's Qo S attribute,such as by real-value(such as price),by interval value(such as reliability),and by natural language(such as credibility).Based on fuzzy mathematics theory,this thesis uses triangular fuzzy number to normalize Qo S attributes,and gives the representation method of composite service's Qo S aggregation function and the calculation method of fitness function.The triangular fuzzy number can be used to enable more accurate representation and analysis of the Web service's Qo S attributes,which lays a theoretical foundation for the research on service combination in the following chapters.(3)Due to the exponential growth of Web services,many existing Web service composition algorithms are computationally expensive and inefficient,therefore they are not suitable for industrial application.This thesis analyzes and compares existing clustering algorithms,and proposes an ISOM(Improved Self-Organizing Maps)algorithm.It is used to cluster the set of abstract services which is represented by triangular fuzzy numbers.This method can greatly decrease the solution space and reduces the overhead of the service composition algorithm.Experimental results show that the ISOM algorithm has better performance.(4)After reducing the space of problem solution,it is necessary to design an efficient service combination algorithm.This thesis not only considers the process of service composition and the global restriction attributes,but also considers the conflicts and dependencies between services.A new Web service composition algorithm,namely C-CGA algorithm(Constrained Chaotic Genetic Algorithm)is designed and implemented.The basic idea of C-CGA algorithm is as follow: Firstly,the ISOM algorithm is used to determine the problem solution space.Then the chaos principle is used to optimize the genetic algorithm,which can be removed the local optimum,improved the convergence speed of the algorithm.Experimental results show that the C-CGA algorithm has a good convergence speed.