Federal element learning-based mobile edge computing intelligent unloading method and device

Abstract

The invention discloses an intelligent unloading method and device for mobile edge computing based on federal element learning. A cloud server and an edge server have neural network models with the same structure. The edge server downloads initial network parameters of the neural network model from the cloud server to update network parameters of a local neural network model, trains the local neural network model, calculates a loss value and uploads the loss value to the cloud server; and the cloud server aggregates all the received loss values to update network parameters to complete training of a network model, and the edge server determines an optimal unloading strategy by adopting a trained neural network model. According to the method, on the premise that user data privacy is not leaked, the multiple edge servers are combined for joint training and learning, the neural network model with the higher generalization ability is obtained, and personalized calculation unloading application of the edge servers is achieved.

Description

technical field [0001] The present application belongs to the technical field of computing offloading of mobile edge computing, and specifically relates to a method and device for intelligent offloading of mobile edge computing based on federated meta-learning. Background technique [0002] With the rapid development of IoT services, it brings a lot of resource requirements for mobile applications (for example, real-time interactive online games and augmented / virtual reality). However, due to the limited computing resources of traditional IoT devices, resulting in reduced quality of experience (eg, long delays) when performing computationally intensive tasks. At the same time, traditional IoT devices are sensitive to energy consumption, so when computing tasks become more and more heavy, energy consumption becomes a major challenge. Mobile edge computing (MEC) can migrate intensive computing tasks from smart devices to nearby edge servers with sufficient computing resources...

AI Technical Summary

Problems solved by technology

[0004] The purpose of this application is to provide a method and device for intelligent offloading of mobile edge computing based on federated meta-learning, so as to avoid the problem that existing technical solutions cannot meet the QoS requirements of different mobile terminals in a dynamically changing network scenario

Images