Adaptive Task Allocation for P2P-Based e-Services Composition

To effectively manage the task allocation, especially when handling with numerous different peers' qualities, is one of the greatest challenges to be faced in order to guarantee the success of P2P-based e-services composition. In this context, various QoS descriptive frameworks and Web services technologies (such as WSDL and BPEL) are being considered as the most affordable solutions to promote the performance of decentralized e-services, through applying strategies like QoS ontologies and related optimization algorithms globally or locally. Nonetheless, most P2P-based service selection and composition approaches applied nowadays lack dynamism and autonomy. In this paper, we first propose an extension of non-functional properties in WSMO, so that to globally facilitate dynamism and autonomous coordination in service compositions. Furthermore, taking into account a model driven approach, we design a planning algorithm to intelligently assign composition tasks to the most appropriate peers for different steps in a whole process. This algorithm is implemented in our prototype UOW-SWS via considering a typical LoanApproval scenario.