Federated Learning Terminal Selection via Mobile Core Network Feedback
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current federated learning methods face challenges in selecting participating terminals efficiently due to inaccurate terminal and network performance information, leading to inefficiencies, errors, and increased time and cost, especially when dealing with a large number of terminals with varying network and computational capabilities.
Innovation Solution
A federated learning method that utilizes a mobile core system to query and integrate terminal and network performance information using AI support functions (AISF), enabling accurate selection of participating terminals and optimizing parameter transmission through dedicated routes and slices.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If terminal information and network performance information are obtained through indirect application layer measurement, then the system can operate without centralized data collection, but the information accuracy is insufficient leading to errors in terminal selection
Solution Approach 1:
The patent introduces the mobile core network as an intermediary between terminals and the federated learning system. The mobile core network collects and manages accurate terminal information and network performance information centrally, then provides this information to the federated learning system via the NEF. This intermediary approach maintains the distributed nature of federated learning while eliminating the need for indirect measurement, thus improving both reliability and measurement precision simultaneously.
2Productivity
If all terminals are allowed to participate in learning iterations, then learning coverage is maximized, but network resources are excessively consumed and errors increase due to including unstable terminals
Solution Approach 1:
The patent implements a feedback mechanism where the mobile core network continuously monitors terminal status and network performance, then provides updated information to the federated learning system. This feedback loop enables the system to dynamically adjust terminal participation based on current conditions, excluding unstable terminals from learning iterations while maintaining high learning efficiency through accurate, real-time information about terminal reliability.
3Adaptability or versatility
If terminal information is determined independently through other methods, then system autonomy is maintained, but users have to take risks of errors occurring and additional time and cost are required
Solution Approach 1:
The patent applies preliminary action by having the mobile core network pre-collect and organize terminal information and network performance information before federated learning iterations begin. The information is prepared in advance and made available through the NEF when needed, eliminating the need for time-consuming independent determination during learning operations. This preliminary preparation maintains system autonomy while significantly reducing the time and computational resources required during actual learning iterations.
Data Source
AI summary
The present invention related to a method for federated learning method interworking with a mobile core system, the method comprising: querying terminal information of each individual terminal among a plurality of terminals; querying network performance information; selecting participating terminals among the plurality of terminals on the basis of the terminal information and the network performance information; transmitting respective parameters to the participating terminals and requesting local learning; and integrating the parameters.