Application of deep learning based on cloud edge

• 1、School of Information and Communication Engineering, Beijing University of Posts and

Abstract：In order to solve the network performance and privacy problems of the deep learning model of the Internet of things, this study proposes to apply the deep learning algorithm to the cloud side collaborative system to optimize the network performance and protect the user privacy in data upload. The multi-layer structure of deep learning is applicable to the cloud side collaboration system. The edge nodes upload reduced intermediate data, so the network traffic from the edge side devices to the cloud server is reduced. Considering the limited service capacity of edge nodes and the real-time fluctuation of network bandwidth, an adaptive algorithm for computing task unloading in cloud edge collaborative computing environment is designed to optimize the overall application performance of cloud edge collaborative system. Experimental results show that the algorithm can dynamically adjust the computing task unloading strategy according to different network environments and server performance in the cloud side collaborative computing environment, and improve the response speed and robustness of the cloud side collaborative system.

Keywords： Internet of things Cloud edge collaboration system Deep learning Adaptive algorithm for computing task unloading