| As the Web Services technology is more and more mature in the field of distributed services provided, the behavior control of Web Service Endpoint(WSE) was more and more important. With the increasing of system complexity, it’s more and more difficult for controlling the behaviors of the server. So how to make WSE can self-controlling had become a hot topic. This article use neuro-fuzzy theory to designing the WSE, and discussed how to adjust the neural network through online learning which can make the WSE have self-learning ability.First, this article describes the basic theory of fuzzy control and neural network. And then, combination of two intelligent control theory by using neural to training the membership function of fuzzy system. Because of the black-box characteristics of the neural network, we do not need pay too much attention to the design of the membership function. At the same time because of the characteristics of parallel computing, even if the membership function is more complex, it also could quickly calculate the results.After established the basis structure of the neural and fuzzy system, this paper discussion about what kind of control behavior should self-learning controller do for different roles of the WSE (service requester or service provider). For the services provider role, it should determine what strategy should web service endpoint do while a new service request coming. And for the role of the service requestor client controller, the controller’s must exchange information with the server controller, and change the call of the services mode (synchronous or asynchronous).Finally, we build a simple Web Services platform to verify the controllers which discussed in the paper. Through the experiment it can be found that the controler which based on neuro-fuzzy was able to spontaneously adjust the behavior of the web service endpoint, and have a certain degree of self-learning ability. |
