Research And Implementation Of Web Service Quality Deviation Monitoring Method Based On QoS Randomness

Web service increases rapidly as the increasing demands of users. In order to satisfy the changing needs of users, service providers assemble the Web services to realize more complex functions. Quality of Service (QoS) as an important standard to measure the quality of Web service has become a significant criterion for service users to select a Web service. QoS attribute of the Web service usually refers to throughput, success rate, response time, availability, reliability, applicability, etc.The Web service runtime environment is highly open and dynamic which will inevitably makes the QoS have intrinsic randomness. In addition, there is some QoS randomness caused by the external human factors. The human here mainly refer to Web service providers and users. In order to improve the chance of their services being selected by users, inevitably there will be some fraud behavior that the QoS of their providing Web service in different stages may be inconsistent. Service customers will get the asymmetry information, so that they can’t obtain the real QoS information for choosing the services. They will face the risk can not getting their really need services. Some service users will deliberately belittle the real QoS level for their private benefits.We need to monitor the quality of Web service to reduce the risk of Web service QoS. In view of the above circumstances, after analysing the existing domestic and foreign research about random QoS aspects of Web services, we designe a Web service QoS deviation monitoring method based on random QoS. Firstly, we regard the QoS of Web service as random variable and utilize distribution function to describe the stochastic probability model. Then, the deviation of random QoS is defined. Secondly, we use Kolmogorov-Smirnov (K-S) test method to test the difference of Web service QoS in different stages and utilize stochastic dominance decision approach to judge the deviation among them. Finally, we carry out experiments on real data sets, the results show that the designed algorithm can effectively detecte the QoS deviation of Web services in different stages, and it is verified that the running time of the method can terminate in acceptable time with the sample volume growth.