Task unloading system and method based on machine learning

Abstract

The invention relates to a task unloading system and method based on machine learning, and belongs to the technical field of edge computing task unloading. The unloading system comprises an agent control module, an unloading decision module, a task classification module, a data management module and a local execution module, wherein the unloading decision module comprises an unloading prediction unit and a data analysis unit. According to the invention, the nearest edge server is connected through the agent control module; for the executed unloading tasks, the task classification module uses adictionary tree to classify the tasks; for the divided sub-tasks, an unloading prediction unit gives an unloading prediction; the data analysis unit counts unloading results and records unloading execution data at the same time; the local execution module is responsible for executing the sub-tasks of which the unloading results are locally executed; the data management module is responsible for data transmission between the mobile device and the edge server. The unloading decision prediction accuracy is high, and the unloading efficiency can be improved.

Description

technical field [0001] The invention relates to a machine learning-based task offloading system and offloading method, belonging to the technical field of edge computing task offloading. Background technique [0002] Edge computing is a small data center that is geographically closer to users and terminal devices than cloud data centers. Edge computing provides users with services closer to the data source on the premise of ensuring low latency, and at the same time solves the problems of low real-time performance, poor privacy protection, and high energy consumption under the cloud computing model. The purpose of edge computing is to make up for the shortcomings and deficiencies of cloud computing, and to offload some tasks that were originally offloaded to the cloud data center to the edge data center for execution. The quality of the uninstallation decision directly affects the execution efficiency of the application, which in turn affects the user experience. [0003] ...

AI Technical Summary

Problems solved by technology

[0004] Aiming at the problem that task offloading by mobile devices cannot be actively predicted and the delay and energy consumption of task offloading, we propose a machine learning-based task offloading system and offloading method, which can comprehensively offload task types, network delays, offloading Task data, mobile device energy consumption and other conditions, use support vector machine to predict the offloading process, and make task offloading decisions based on the prediction results

Images