Research On Web Services Composition With Temporal Constraint

The Service Oriented Architecture(SOA)based on Web services is becoming a mainstream software paradigm on the Internet.Through reusing and assembling of the existing Web services,the flexible integration of data,resources and systems can be realized.With the proliferation of Web services available online,optimization of compositing Web services under the constraint of timing and Quality of Services(QoSs)is necessary.Majority of current research is focused on temporal compatibility verification or calculus of composite services.And QoS-aware Web services composition is still plagued because of high time overhead,especially when uncertainty QoS is taking into account.Based on the above backgrounds,this dissertation focuses on the research of Web services composition with timing and QoS(e.g.the response time)constraints,and obtains the following results.(1)A temporal constraint decomposition model and its algorithm are proposed.The goal of this model is to transfer the global constraints applied to the total workflow or sub workflows into a serial of local constraints applied to each task.Temporal constraints specified by the user will be assured when selected candidate service for each task satisfies their local temporal constraints.The greedy algorithm will retain as many composite services candidate as possible.That is,the algorithm will relax the local constraints as much as possible,as well as balance the number of reserved services for each task.Through the process of temporal constraints decomposition,the problem of Web service composition based on temporal constraints is transformed into a general Web service combination problem.In addition,those services which violate local temporal constraints will be filtered out.Thus,the scale of the original problem is reduced to some extent.The experiments demonstrate the effectiveness,necessity and time overhead of the proposed method.A performance evaluation is also established based on the comparison with previous work.(2)Although TCD model can safeguard global temporal constraints,it may lose feasible solutions.Therefore,when the user constraint intensity is strong,the probability of finding a feasible combination is low.In order to solve this proble:m,relaxation factor is introduced into TCD model to form CIA-TCD model.And an adaptive adjustment algorithm based on fuzzy reasoning is presented for relaxation factor.Simulation results show that,compared with TCD model,CIA-TCD model can significantly improve the probability of finding a feasible combination scheme when the constraint strength is strong.(3)A two-stage Web service combination approach based on global QoS constraints decomposition is introduced.During the first stage,the global constraints are decomposed into a set of local constraints that serve as the conservative upper/lower bounds for each task.The purpose of this decomposition model is to maximize the number of reserved services and guarantee the global QoS constraints.As a solution of this model,the greedy algorithm has low time complexity,and it has a low correlation to the values of QoS,so it allows the separation of the constraints decomposition process and local selection process.The second stage is a local optimization,which only needs to pick the service candidate with optimal utility and met local constraints obtained from the first stage.Therefore,the proposed approach is more suitable to a dynamic environment.By Comparing to previous work,the experimental results indicate that our approach improved utility and reduced time complexity.(4)To describe the problem of Web service composition with uncertain QoS,an Integer Quadratically Constrained Program(IQCP)model is proposed This paper is focused on workflow presented by a Directed Acyclic Graph(DAG).By assuming that QoS to a'Web service follows a normal distribution and different Web services are mutually independent,four types of expectation and variance calculation are discussed.And then,Web services composition with uncertain QoS is assembled as an IQCP model.Corresponding robust and time overhead are analyzed through CPLEX,MATLAB,and JAVA.The results also show that this approach is faster and more accurate compared to previous approaches(5)After introducing a temporal model of Web service with available time interval restriction,a Web service composition method is presented to minimize the makespan.The experiments analyze time overhead of this method and verify the necessity of studying this problemIn conclusion,this dissertation conducts an in-depth study on the problem of Web services composition related to timing constraints.Through the models of temporal constraints decomposition and global QoS constraints decomposition,the temporal and QoS constraints can be decomposed into a set of local constraints.Thus,a composite service assures user requirements can be quickly found by local optimization.The proposed approaches have lower complexity and more suitable.for dynamic scenarnos.Therefore,it has certain academic values and can be used to inspire future studies.This dissertation also proposes an IQCP model to describe Web services composition with uncertain QoS and explores the problem of Web services composition under available time interval restriction.The related work needs to be improved in future.