MDP Planning Of Web Services Composition With Time And Resource Constraints

If no single Web service can satisfy the functionality required by the user, there would be a possibility to combine existing services together in order to fulfill the request. This trend has triggered Web services composition. The network technology is developing rapidly and the process of global economic integration is accelerating, which make both the internal and external situations of enterprise change enormously. And an increasing amount of companies and organizations only implement their core business and outsource other application services over Internet. Thus, the ability to efficiently and effectively select and integrate inter-organizational and heterogeneous service on the Web at runtime is important step towards the development of the Web service applications.Classical planning suffers from the paradox of first assuming deterministic behavior of Web service, then requiring the additional overhead of execution monitoring to recover from unexpected behavior of services. The indeterminate problem of web service composition can be solved by MDP(Markov Decision Process). But the recent MDP frameworks suffer from some limitations: They are failing to consider about time and resource constraints. Otherwise, the time and resource consumes of Web service are uncertainly.The markov decision process is the proper method to solve such problems. We can calculate the interval and source level for each Web service offline to get a fully observed states space with time and resource constraints. Furthermore, the paper defined the action as a composition of a task and a task start time. The action can represent how the actions affect the state on time and resource. And we get a transmit matrix of MDP. At last, we define a reward function which considers the time and resource consume.In the end of the paper, some web service composition cases were modeled and them were solved by some classical MDP arithmetic. The dates represent the framework has the ability to get optimal policy on the problem. The results prove that framework can handle the web service composition with time and resource constraints usefully.