Research On Adaptive Prediction Method Of Web Service QoS

The quality of service(QoS) of the web service has a strong dynamism due to the complicated and changeable environment such as the openness of the network environment in which web service is located, the fluctuation of service loads, and the uncertainty of the use’s needs. For the accuracy of the service QoS is the main factor that affects the success rate of service selection and service composition, how to accurately predict the QoS of web service has become a key scientific issue that needs to be solved urgently in the field of services computing and cloud computing. Because the diversity of the service scene that users’ use, the different rich degrees of the Web service QoS data and the limitation of QoS prediction methods, it is difficult to find a QoS prediction method that can accurately forecast the QoS in a frame that includes various scenarios.In view of the above issue, this paper constructs a Web service QoS adaptive prediction model(WS-QoSAPM). The model divides the QoS prediction into real-time QoS prediction and the QoS prediction after a period of time, and adapts different QoS prediction methods respectively in different scenarios. The main works of this dissertation are outlined as follows:1. In the real-time prediction, the paper achieves the Web service QoS prediction based on two methods: the optimized support vector machine based on the improved artificial bee colony and the case-based reasoning. According to the experimental results,the paper formulates the corresponding adaptive prediction strategy: when the QoS data is inadequate, use the method of CBR to predict the QoS; When the QoS data is adequate, use the method of O-SVM.2. For the prediction of QoS after a period of time, the paper proposes the method of O-SVM+CBR and O-SVM+O-SVM. In the method of O-SVM+CBR, use the O-SVM to predict the load, then based on the CBR to predict the QoS. In the method of O-SVM+O-SVM, it also use O-SVM to predict the load, then based on the O-SVM to predict the QoS. At last, the paper formulates the corresponding adaptive prediction strategy through the experiment result: when the QoS data is inadequate, using the O-SVM+CBR to predict the QoS; When the QoS data is adequate, using the method of O-SVM+O-SVM.3. According to the above research results, the paper proposes the WS-QoSAPM.The model can choose the optimal prediction method based on the character of QoS data in different application scenarios.The proposed adaptive QoS prediction method can choose the most suitable QoS prediction method for the service in different application scenarios, which improve the efficiency of the web service QoS prediction in various scenarios, and provide effective support for Web services selection with QoS aware service composition.