DOI: 10.14704/nq.2022.20.8.NQ44721

Web Service Composition using Markov Decision Process and Long Short Term Memory

Subbulakshmi S,Hemant P Hareesh, Rahul R

Abstract

Software and Systems are abstracted as Web Services in the Service-Oriented based design, which may then be used by other systems. Using the approach of service composition, simple services that already exist can be combined to produce sophisticated solutions. Massive web services with the same functionality start to emerge as web service technology develops. Numerous groups maintain these services, and the level of service varies. As a result, a crucial problem in service composition research is how to pick the optimal service to guarantee that the entire system gives the best overall QOS (Quality Of Service). Additionally, Given the complex nature and fluctuation of the network environment, QOS may alter over time. Consequently, it is difficult to comprehend how to dynamically modify the composition system to react to shifting settings while keeping the calibre of the composing service. To get over the current issues, we propose a method for composing services that is based on QOS prediction and reinforcement learning. After employing a Long Short-Term Memory (LSTM)which is a Recurrent Neural Network which is used to predict QOS, we use reinforcement learning to select dynamic services. Our approach is easily adaptable to a dynamic network setting. We put our plan through a number of tests to make sure it works.

Keywords

Deep Learning, Long Short Term Memory, Recurrent Neural Networks, Quality of Services, Web Service Composition

Full Text

References