VFL Client Alignment for 5GC Analytics Model Training
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing federated learning systems in wireless communication networks face challenges in efficiently aligning samples and features across multiple network elements for machine learning model training, leading to suboptimal performance and inefficiencies in network data analytics.
Innovation Solution
A method and system for vertical federated learning (VFL) that involves receiving information about network elements and their ML models, performing sample and feature alignment, and determining a global and local ML models for analytics services, with capabilities to update and communicate training results across VFL clients.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If federated learning is performed across multiple network elements, then model training capability is improved, but sample and feature alignment complexity increases
Solution Approach 1:
The patent introduces a federated learning manager as an intermediary component that coordinates between multiple network elements. This manager handles the complex tasks of sample alignment and feature matching, centralizing the alignment logic and reducing the complexity burden on individual network elements while enabling collaborative model training across the network.
Solution Approach 2:
The patent performs sample alignment and feature matching as preliminary actions before the actual model training process. By pre-aligning samples and features across network elements and establishing correspondence relationships in advance, the system reduces the complexity during the training phase and enables more efficient federated learning execution.
2Productivity
If vertical federated learning is implemented with multiple VFL clients, then analytics service performance is improved, but communication overhead increases
Solution Approach 1:
The patent extracts and separates the alignment results and model parameters from the raw data at each network element. By taking out only the necessary alignment information and training updates rather than exchanging complete datasets, the system reduces communication overhead while maintaining analytics service performance across multiple VFL clients.
Solution Approach 2:
The patent implements partial action by performing alignment and training operations selectively based on data availability and client capabilities. The system determines which samples and features to align and which clients to involve in each training round, avoiding unnecessary communication and computation while still achieving improved analytics performance through coordinated multi-client participation.
Data Source
AI summary
An example method performed by a vertical federated learning (VFL) server is disclosed. The method comprises receiving information indicating a plurality of network elements associated with one or more analytics services and respective ML models supported at the plurality of network elements for the one or more analytics services, selecting one or more of the plurality of network elements as one or more VFL clients to perform VFL for an analytics service based on the received information, performing sample and feature alignment with the one or more VFL clients to obtain alignment results, and determining a VFL model for the analytics service. The VFL model includes a global ML model for the VFL server and one or more local ML models for the one or more VFL clients. The method further comprises sending an indication of a local ML model determined for each VFL client.