Federated Learning for NWDAF Model Training
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Collating data at a single Network Data Analytics Function (NWDAF) in wireless communications networks is slow and resource-intensive, and may pose security risks, especially when handling sensitive or private data.
Innovation Solution
Implementing federated learning across multiple NWDAFs, where each local NWDAF stores and trains a copy of the model using machine-learning, with a central NWDAF combining local model updates to obtain a combined model update, thereby reducing resource overhead and security risks.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If data is collated at a single NWDAF, then centralized model training can be performed, but the process becomes slow and resource-intensive
Solution Approach 1:
The patent segments the centralized model training process into multiple distributed training operations performed by different NWDAF instances. Each NWDAF trains a local copy of the model using its own data, and the results are aggregated to form the final model. This segmentation distributes the computational workload across multiple nodes, improving training speed while maintaining model accuracy through collaborative learning.
2Loss of information
If data is collated at a single NWDAF, then comprehensive data analysis can be performed, but security risks increase when handling sensitive data
Solution Approach 1:
The patent extracts the sensitive training data from the centralized NWDAF and keeps it distributed at local NWDAF instances. Only the trained model parameters and performance metrics are aggregated, not the raw sensitive data itself. This extraction approach maintains complete data utilization for model training while eliminating security risks associated with centralizing sensitive information.
3Productivity
If multiple NWDAFs are used for model training, then resource overhead is reduced, but coordinating model updates becomes complex
Solution Approach 1:
The patent merges the model updates from multiple NWDAFs through a standardized aggregation process. Each NWDAF contributes its trained model parameters, and these are combined using predefined rules (such as averaging or weighted aggregation) to produce a unified model update. This merging approach reduces resource overhead by distributing training work while managing coordination complexity through systematic aggregation protocols.
Data Source
AI summary
Aspects of the disclosure provide a method performed by a central Network Data Analytics Function (NWDAF) in a communications network. The communications network comprises one or more local NWDAFs configured to develop a model using federated learning, in which each local NWDAF stores a copy of the model and trains the copy of the model by inputting training data into a machine-learning process. The method comprises receiving, from the one or more local NWDAFs, a respective local model update comprising an update to values of one or more parameters of the model generated by training a respective copy of the model using machine-learning. The method further comprises combining the local model updates received from the one or more local NWDAFs to obtain a combined model update.