Deep Learning-based QoS Prediction for Manufacturing Cloud Service

Multiple manufacturing cloud services (MCSs) are integrated in cloud manufacturing platform for providing service to internet users and its QoS has become an important evaluation indicator. Availability and reliability are two important properties of QoS. But few researches have been done on availability prediction and MCSs are always supposed to be available, while reliability is usually estimated by the empirical value or the mean value of historical executions. However, they both considered a little or even ignored the dynamic characteristics of cloud environment. This paper designed a deep learning based approach to predict QoS, i.e. availability and reliability, where availability prediction utilizes LSTM, and reliability prediction uses DNN model. To validate the effectiveness of the proposed method, the experiment is conducted and its results demonstrate that our approach outperforms the existing ones.

1 Introduction

Nowadays, product personalization, manufacturing globalization and green manufacturing pose great challenges to enterprises. In this context, Cloud Manufacturing (CMfg) develops rapidly, based on Cloud Computing, industrial Internet of Things, autonomous perception, big data as well as artificial intelligence [1–2]. More and more manufacturing cloud services (MCSs) are registered in manufacturing cloud platform, which can be used to transparently provide users with reliable, superior and on-demand services to meet QoS[3].