5G • 5G-Advanced (Rel-18)
AI/ML Lifecycle Management (LCM)
3GPP Air Interface AI/ML Model Lifecycle Management
Standardized signaling framework for deploying, monitoring, retraining, switching, and falling back AI/ML models across the UE and gNodeB.
Technical Explanation
AI models running on cellular air interfaces must adapt to non-stationary radio environments and hardware differences across UEs. Release 18 establishes AI/ML Lifecycle Management (LCM), specifying standard RRC and MAC CE protocols for model capability negotiation, model selection, performance degradation monitoring (concept drift), model activation/deactivation, and deterministic fallback to legacy 3GPP non-AI mechanisms.
Key Functions
- Defines RRC signaling for UE AI/ML model capabilities and compatibility matrix
- Monitors inference accuracy and triggers model updates when drift is detected
- Executes seamless switching between localized baseband neural network models
- Provides failsafe fallback to standard 3GPP Release 15-17 non-AI algorithms
Specifications
3GPP TR 38.843, TS 38.331 (Rel-18)
Interfaces
Uu (RRC)MAC CEO1
Related 5G Concepts
Want to memorize 5G concepts like this one?
Study it with SuperMemo SM-2 spaced repetition flashcards.