A Deep Learning Based Efficient Data Transmission for Industrial Cloud-Edge Collaboration

Abstract

The tremendous data transmission between the cloud server and edge gateways accelerates the realization of the intelligent factory. However, it consumes enormous band-width resources and leads to the problem that limited factory bandwidth can not meet the large-scale high-density online data transmission. Therefore, data transmission between the cloud server and edge gateways must be reduced to enable large-scale cloud-edge interaction. To achieve this purpose, we propose a deep learning (DL) based data transmission reduction (DPTR) scheme for cloud-edge collaboration, which combines the cloud-edge characteristics to reduce the data transmission volume online while ensuring data accuracy. Meanwhile, we built a physical verification platform including sensor, edge gateway, and cloud server to collect real data and validate the DPTR scheme. Based on the physical validation platform and real data, we experimentally demonstrate that the proposed scheme can reduce the data transmission by 76.83 % while guaranteeing the relative deviation of less than 10%, even for drastically changing vibration data.