Online Reliability Prediction Via Dae And Heuristic Algorithm For Service Systems

The composited service system based on SOA integrates various Web services with different functions to satisfy users' changeable and complex needs.These selected component services issued by third parties usually have a bad performance,which will eventually affect the performance of the composited service system.This makes the composited service system constructed by these component services unable to meet the needs of users.Therefore,an effective measures must be taken to ensure performance of the composited service system.To keep the performance of software,many works use the proactive fault management to adjust the component of system by predicting which component will crush.This method has been proved to be an effective way to ensure the stablility of the system.The composited service system can also be used in this way,and then the problem that how to ensure that the stability of the system will be transformed into how to monitor the performance of each component service in a composited service system.In this paper,we monitor the performance of component services by predicting its reliability.In this paper,the denoising autoencoder and a heuristic BP algorithm are used to predict the reliability of service.First,the denoising autoencoder is used to extracte the features of the parameter time series of the component service in the window time period.On this basis,the BP algorithm is used to learn the relationship between parameter and the reliability time series in prediction phase.To validate the effectiveness of the model,we conduct a series of experiment to check whether the proposed model is better than others.The experimental results show that the model proposed in this paper has a strong advantage over the reliability prediction compared with other algorithms.