NWDAF ML Model Lifecycle Management for Versioned Analytics
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In 5G mobile communication systems, there is a lack of a systematic approach for managing machine learning (ML) models within network data analytics functions (NWDAF) devices, which hinders their effective utilization and optimization.
Innovation Solution
A method for managing ML models in NWDAF devices through mechanisms such as model discovery, provisioning, updating, and sharing, involving operations like ML model discovery request, registration, subscription, and notification services, with support for local and global training, and consideration of factors like S-NSSAI, Analytic IDs, and serving areas.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If ML models are deployed in NWDAF devices without a systematic management approach, then device complexity is reduced, but the effectiveness and optimization of ML model utilization deteriorates
Solution Approach 1:
The patent segments ML model management into distinct functional modules including model discovery (invoking discovery request/response service operations), model provisioning (subscription and notification service operations), model updating mechanisms, and federation learning coordination. Each module handles specific aspects of ML model lifecycle independently, reducing overall management complexity while maintaining effectiveness.
Solution Approach 2:
The patent introduces an intermediary management framework that mediates between ML model sources (including external providers and federation learning participants) and NWDAF devices. This intermediary layer handles model registration, discovery, provisioning, and updating through standardized service operations, simplifying the utilization process while enhancing effectiveness.
2Adaptability or versatility
If multiple ML model versions are managed without a systematic approach, then model adaptability is reduced, but information management complexity deteriorates
Solution Approach 1:
The patent implements dynamic ML model version management where the NWDAF device can subscribe to and receive notifications about multiple model versions. The system dynamically updates models based on notifications from providers or federation learning outcomes, allowing adaptability across different versions without manual intervention or complex version control mechanisms.
Solution Approach 2:
The patent establishes feedback loops through notification service operations where model providers or federation learning participants notify the NWDAF device about model updates. This feedback mechanism enables automatic model version updates, maintaining adaptability while simplifying information management through event-driven updates rather than continuous monitoring.
Data Source
AI summary
A machine learning (ML) model management method for a network data analytics function (NWDAF) device is disclosed. The NWDAF device performs at least one of an analytics logical function (AnLF) for network data and an ML model training logical function (MTLF).