Research On Web Service Discovery And Selection Basing On P2P Network

In recent years, as a new distributed computing and software application paradigm, services computing has attracted extensive attention and research. The kernel thoughts of services computing are software reuse and loose coupling. It bases on SOA(Service-Oriented Architecture), and takes web services as functional modules of software. Briefly, SOA is a software architecture in which functional modules in a software is assembled by pre-defined interfaces and contracts. Such modules are called "Services" in SOA and they are loosely coupled. With the popularity of Internet and extensive usage of XML, Web service becomes the main implementation technique for SOA. W3C defines a web service as "a software system designed to support interoperable machine-to-machine interaction over a network". When developing a software application, after analyzing the functional requirements of the application, developers could take published services to meet the requirements and assemble services to finish the development. The quickness and low cost of development by SOA fit the diversification of user requirements and the customization of software applications well.With the fast development and extensive application of web services, there are more and more service providers providing more and more web services. The need for service registration and discovery is increasing day after day. Traditional service discovery technique which bases on UDDI registry is centralized. Its reliability and performance can not meet the increasing requirement of service discovery. At the same time, with the increasement of services, there would be many services could satisfy certain functional service requirement. Therefore selecting a best service according to quality of Services becomes an important research issue. Traditional QoS-aware service selection methods all assume they are working on the whole candidate service set, so many of them can not get correct results when working on distributed service set.Based on previous works, this paper explores above issues and proposes corresponding solutions. Firstly, we analyze the drawbacks of centralized service discovery pattern, and proposes a new service registration and discovery method basing on unstructured P2P network which is more reliable than traditional method. Then, considering the new P2P service discovery environment, we analyze the properties of service selection algorithms which can get correct result in such environment. Finally, we propose an algorithm using skyline to select services, and prove the algorithm satisfies above properties.