AI/ML Model Lifecycle Control in Mobile Communication
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing mobile communication systems face challenges in effectively managing the life cycle operations of artificial intelligence (AI)/machine learning (ML) models, such as selection, activation, deactivation, switching, and updating, which are crucial for optimizing wireless communication performance.
Innovation Solution
A communication control method is introduced that enables user equipment to manage the life cycle operations of AI/ML models by transmitting life cycle management (LCM) operation disable information and execution condition information, allowing for efficient switching between training and inference modes to optimize resource usage and reduce overhead.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If life cycle management operations for AI/ML models are implemented in mobile communication systems, then model selection, activation, deactivation, switching, and updating capabilities are improved, but system complexity and overhead increase
Solution Approach 1:
The patent segments the life cycle management of AI/ML models into distinct operational phases (selection, activation, deactivation, switching, updating) that can be independently controlled and managed. Each phase can be enabled or disabled separately through configuration parameters, allowing the system to implement only the necessary management operations rather than handling all possible operations simultaneously, thus reducing overall system complexity while maintaining comprehensive model management capability.
Solution Approach 2:
The patent implements dynamic control of LCM operations through configuration parameters that can be adjusted based on system conditions, network state, and operational requirements. The system can dynamically enable or disable specific LCM operations (selection, activation, deactivation, switching, updating) and transition between different operational modes, allowing flexible adaptation without permanently increasing system complexity.
2Productivity
If AI/ML models are deployed for wireless communication optimization, then communication performance is improved, but power consumption increases
Solution Approach 1:
The patent implements periodic or conditional execution of AI/ML model operations rather than continuous operation. The system can periodically evaluate whether model operations are necessary based on changing communication conditions, and can suspend or deactivate models when they are not needed, thereby maintaining communication performance when required while reducing power consumption during normal operation.
Solution Approach 2:
The patent enables the system to discard (deactivate or remove) AI/ML models when they are no longer needed for optimization, and recover (reactivate or reload) them when conditions warrant their use. This selective activation and deactivation of models allows the system to achieve performance improvements only when necessary, significantly reducing overall power consumption while maintaining productivity when models are active.
Data Source
AI summary
The present disclosure relates to a communication control method in a mobile communication system. In the communication control method, a model transmission entity transmits, to a model reception entity, life cycle management (LCM) operation disable information indicating whether a life cycle management (LCM)-related operation is disabled for an artificial intelligence (AI)/machine learning (ML) model and/or LCM operation condition information indicating an execution condition for the life cycle management (LCM)-related operation for the AI/ML model. Here, the life cycle management (LCM) related operation is any one of selection of an AI/ML model, activation of the AI/ML model, deactivation of the AI/ML model, switching of the AI/ML model, fallback from the AI/ML model to a non-AI/ML model, update of the AI/ML model, or transfer of the AI/ML model.