Research On Reliability Prediction Of Service-Oriented Systems Based On Critical Component Identification

The service-oriented systems based on Service-Oriented Architecture(SOA)can integrate services to meet the needs of users.However,The service-oriented systems are defined as loosely coupled software applications,which run in a dynamically changing network environment.And their reliability will fluctuate due to external reasons such as location,communication link and internal reasons of the service.The reliability prediction of service-oriented systems can know the reliability states in advance,so as to prevent the occurrence of emergencies and ensure the stable operation of service-oriented systems.In the field of service computing,the reliability prediction of service-oriented systems has become a research hot spot.At present,reliability prediction methods of service-oriented systems are mostly studied to improve the accuracy of prediction,but the timeliness of reliability prediction is ignored.The purpose of this paper is to improve the reliability prediction accuracy and the prediction efficiency of service-oriented systems,so as to ensure the stability of service-oriented systems.Based on the analysis and summary of the reliability prediction methods of service-oriented systems,this paper proposes a reliability prediction framework for service-oriented systems based on critical components identification.The critical component identification method and reliability prediction method of service-oriented systems are studied.A critical component identification method based on improved weighted LeaderRank algorithm is proposed.In this method,the critical components are identified according to the importance of service-oriented systems components by considering the factors of service dependency,service popularity and service propagation ability,which improves the accuracy of component identification.In the process of reliability prediction of service-oriented systems,a reliability prediction model for service-oriented systems based on Attention-LSTM neural network is proposed firstly.This model uses attention mechanism to optimize LSTM neural network and improves the accuracy of reliability prediction of critical components.Then,the critical service composition model graph of service-oriented systems is constructed based on the depth first traversal algorithm to describe the composition structure among the critical components.Next,a reliability calculation method for service-oriented systems is proposed,which can obtain the reliability prediction value of service-oriented systems by using the reliability prediction value of critical components.Finally,the reliability prediction framework for service-oriented systems based on critical components identification is implemented.Experiment results show that the proposed method in this paper has obvious advantages in accuracy and efficiency of service-oriented systems reliability prediction.