UE ML Capability Exchange for RRM Authorization and Context Alignment
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing 5G networks lack efficient mechanisms for dynamically managing and authorizing Machine Learning (ML) capabilities in User Equipment (UE) for Radio Resource Management (RRM), particularly in scenarios involving ML model lifecycle changes and decoupling UE capabilities from RAN capabilities.
Innovation Solution
A method for enhancing UE capability retrieval by associating ML models with RRM functions and contexts, enabling dynamic notification and authorization of ML models for specific RRM contexts, and supporting ML capability exchange between network entities and UE.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If ML capability exchange mechanisms are implemented in 5G networks, then RRM efficiency and adaptability are improved, but system complexity increases due to additional signaling and authorization procedures
Solution Approach 1:
The patent segments the ML capability exchange by introducing specific information elements (IEs) for ML model identification, capability indication, and authorization. These segmented capability indicators are embedded within existing RRC messaging structures, allowing ML-related information to be extracted and processed independently from general UE capability procedures.
Solution Approach 2:
The patent extends existing 5G RRC signaling structures to serve dual purposes: traditional UE capability reporting and ML-specific capability exchange. The UE Capability Information message and associated IEs are designed to handle both conventional radio capability information and ML model capability information through a unified signaling framework.
2Adaptability or versatility
If dynamic ML model authorization is implemented for RRM functions, then adaptability of ML capabilities is improved, but signaling overhead increases due to frequent capability updates and notifications
Solution Approach 1:
The patent implements event-triggered periodic reporting where the UE monitors ML model status changes (training completion, inference readiness, re-training events) and notifies the network only when state transitions occur. This periodic monitoring mechanism at the UE side replaces continuous network-side polling, reducing uplink signaling while maintaining real-time awareness of ML capability states.
Solution Approach 2:
The patent performs preliminary ML capability indication during initial UE capability reporting, where the UE provides advance information about supported ML models and algorithms. This preliminary exchange allows the network to pre-configure authorization policies and prepare RRM functions before actual ML model deployment, reducing the need for frequent dynamic signaling during operation.
3Ease of operation
If ML capability information is tightly coupled with RAN capabilities, then ease of operation is improved, but flexibility of ML model deployment is reduced
Solution Approach 1:
The patent segments ML capability information from general RAN capability reporting by introducing dedicated ML-specific information elements within the UE Capability Information structure. This segmentation allows ML capabilities to be independently identified, filtered, and authorized without being constrained by traditional RAN capability coupling, enabling flexible ML model deployment while maintaining operational simplicity through structured information separation.
Data Source
AI summary
The present invention provides apparatuses, methods, computer programs, computer program products and computer-readable media for ML capability exchange and authorization for RRM. The method comprises receiving, from a network entity, an enquiry for providing a machine learning, ML, capability of the user equipment, the enquiry including an ML filter, indicating at least one ML algorithm available at the network entity, generating ML capability information indicating at least one ML model available at the user equipment based on the ML filter received from the network entity, and transmitting the generated ML capability information to the network entity.