Adaptive Service Composition Based On QoS Prediction And Multi-agent Reinforcement Learning

A service-oriented architecture(SOA)is a style of software design.In this architecture,software and systems are abstracted as web services to be invoked by other systems.Service composition is a technology which builds a complex system by combining existing simple service.With the development of SOA and web service technology,massive web services with same function begin to spring up.These services are maintained by different organizations and have different QoS(Quality of Service).Thus,how to choose the appropriate service to make the whole system acquired the best QoS has become a key problem in the service composition.On the other hand,because of the complexity and dynamics of the network environment,the quality of service may be change over time.Therefore,how to adjust the combination system dynamically to adapt to the changing environment and ensure the quality of the combined service is also a challenge to be solved urgently.Based on the above challenges,this paper focuses on the problem of adaptive service composition under the dynamic environment,and the following works is completed:(1)Applying the traditional time series prediction method to the field of service computing.And the QoS prediction method based on LSTM(Long Short-Term Memory)is realized.(2)Combining reinforcement learning with QoS prediction and applying it to service composition.(3)Using multi-agent technology to share QoS time series among multiple users.Thus speeding up the learning process and improving the quality of service composition.(4)A series of experiments are conducted to verify the method proposed in this paper,which proves the effectiveness of the method in the dynamic environment.