Machine Learning Computing Migration and Management Based on Edge Computing of Multiple Data Sources in the Internet of Things

Abstract

With the implementation of the concepts of smart city and smart home, the number of user-intelligent terminal devices is increasing. The traditional computing framework cannot meet the increasing data volume and computing needs. Edge computing based on multiple data sources of the Internet of things can not only meet the computing needs of users’ intelligent devices but also reduce energy consumption and user computing waiting time. Therefore, this article puts forward the research on the migration and management of deep reinforcement learning computing based on the edge computing of Internet of things multiple data sources, integrates the deep reinforcement computing technology in the edge computing of Internet of things multiple data sources, and optimizes the edge computing migration scheme and resource allocation management. The test results show that deep reinforcement learning can effectively control the cost of computing migration and enable it to complete computing tasks efficiently while maintaining stable operation. Compared with the traditional enhanced algorithm and the minimum migration scheme, the management model can complete the computing migration task with less energy consumption and shorter average computing waiting time.