AI Model Management Module for Adaptive Network Functions
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The management of massive AI models in network architectures is challenging due to the varying conditions and requirements for different functions, necessitating improved methods for storing, updating, and acquiring these models efficiently.
Innovation Solution
A model management module is introduced to provide services for nodes, enabling operations such as storing, updating, and acquiring AI models, with each model having a unique identifier and context for application in control or user plane functions based on terminal conditions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If a network architecture supports massive AI models for different functions and conditions, then the adaptability and functionality of the network is improved, but the device complexity and difficulty of model management increases
Solution Approach 1:
The patent segments the model management function into a dedicated Model Management Module (MMM) that is separate from the network function instances. This segmentation allows the network to support multiple AI models for different functions without increasing the complexity of individual network nodes, as the MMM handles model storage, selection, and updates centrally.
Solution Approach 2:
The Model Management Module acts as an intermediary between the network function instances and the AI models. It mediates the interaction by providing a standardized interface for model acquisition, storage, and updates, thereby simplifying the management of massive models while maintaining network adaptability.
2Adaptability or versatility
If the network architecture stores and manages multiple AI models for varying conditions, then the functionality for different functions is improved, but the storage requirements and resource consumption increase
Solution Approach 1:
The patent implements preliminary action by pre-storing AI models in the Model Management Module before they are needed. The MMM maintains a repository of pre-processed models that can be quickly deployed to network functions when required, eliminating the need for network nodes to store multiple models locally and reducing overall storage requirements.
Solution Approach 2:
The Model Management Module provides universal functionality for managing all AI models across different network functions. This single multi-functional module handles model storage, selection, and updates for various conditions, replacing the need for each network node to maintain dedicated storage for multiple models.
3Productivity
If the network dynamically updates AI models based on conditions, then the performance optimization is improved, but the update frequency and operational complexity increase
Solution Approach 1:
The patent implements feedback mechanisms where the Model Management Module monitors network conditions and automatically selects or updates AI models based on real-time feedback. This automated feedback loop enables performance optimization without increasing operational complexity, as the MMM handles model updates autonomously based on predefined conditions and performance metrics.
Solution Approach 2:
The Model Management Module performs self-service by automatically managing model updates and selections without requiring manual intervention from network operators. It monitors performance metrics and condition changes, then autonomously deploys appropriate models to network functions, simplifying operations while maintaining optimization.
Data Source
AI summary
A model management method and a communication device are provided. The method includes that: a node obtains a model management service provided by a model management module, the model management service supports the node to perform at least one of the following operations: storing one or more models to the model management module; updating one or more models in the model management module; and obtaining one or more models from the model management module.