Integrating recurrent neural networks and reinforcement learning for dynamic service composition

In the service oriented architecture (SOA), 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 services. With the development of SOA and web service technology, massive web services with the 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 to deliver the best overall QoS has become a key problem in service composition research. Furthermore, because of the complexity and dynamics of the network environment, QoS may change over time. Therefore, how to adjust the composition system dynamically to adapt to the changing environment and ensure the quality of the composed service also poses challenges. To address the above challenges, we propose a service composition approach based on QoS prediction and reinforcement learning. Specifically, we use a recurrent neural network to predict the QoS, and then make dynamic service selection through reinforcement learning. This approach can be well adapted to a dynamic network environment. We carry out a series of experiments to verify the effectiveness of our approach.