Federal learning method and system for network uncertainty perception in mobile edge cloud

Abstract

The invention relates to a federated learning method and system for network uncertainty perception in mobile edge cloud. The method comprises the following steps: based on a federated learning (FL) framework, defining an average quantity of training parameters of each user equipment under the condition that an MEC network is uncertain; determining an average model size factor in each federated learning FL task request process; determining the minimum aggregator number and the maximum aggregator number in each federated learning FL task request process; determining the number of aggregators; constructing an auxiliary graph, and determining a position decision according to the auxiliary graph; based on the position decision, determining the total cost generated in each federated learning FL task request process; and taking the resource capacity of the MEC network as a constraint condition, adjusting the number of aggregators according to the total cost, forming the optimal number of aggregators in each federated learning FL task request process, and optimizing the federated learning FL framework according to the optimal number of aggregators. According to the invention, the aggregators are reasonably distributed, and the implementation cost is minimized under the condition of uncertainty of the MEC network.

Description

technical field [0001] The invention relates to the field of MEC network communication, in particular to a federated learning method and system for network uncertainty perception in a mobile edge cloud. Background technique [0002] With the rapid development of 5G and Mobile Edge Computing (MEC), various artificial intelligence applications, such as augmented reality and smart healthcare, are deployed in MEC networks. Such artificial intelligence applications are constantly generating large amounts of data in the MEC network, which needs to be analyzed to improve its accuracy. Traditional training is performed at a centralized location (i.e., a data center) with a powerful GPU, by sending all data from each user equipment (UserEquipment, UE) to the centralized location. However, sending all raw data, including sensitive data, i.e. images of users' faces, to a centralized location could lead to violations of user privacy. Federated Learning (FL) is considered as a promisin...

AI Technical Summary

Problems solved by technology

[0004] The purpose of the present invention is to provide a federated learning method and system for network uncertainty perception in the mobile edge cloud to solve the unreasonable number of aggregator layouts in the existing MEC network and the inability to achieve minimum network uncertainty under the condition of MEC network uncertainty. cost of implementation

Images