Terminal AI/ML Capability Reporting for Adaptive Model Allocation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The challenge of ensuring the reliability, timeliness, and efficiency of AI/ML operations in 5G and 6G mobile terminals is hindered by limited computing power, storage resources, and battery capacity, as well as the need for real-time AI/ML model acquisition and training, particularly in scenarios like 'AI/ML operation splitting', 'AI/ML model distribution', and 'federated learning'.
Innovation Solution
A method for information reporting that involves terminals sending AI/ML capability information to network devices, allowing the network devices to flexibly allocate AI/ML tasks, models, and training parameters based on the terminal's resources, ensuring efficient utilization of processing capability, storage, and battery resources.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If AI/ML operations are performed in mobile terminals with limited resources, then AI/ML service functionality is improved, but processing capability, storage resources, and battery capacity are insufficient
Solution Approach 1:
The patent segments AI/ML operations into two parts: model training and inference. The terminal performs inference operations locally using stored models, while model training and updates are performed remotely by network devices or servers. This segmentation allows the terminal to provide AI/ML services without requiring full processing and training capabilities on-device.
Solution Approach 2:
The patent introduces network devices as intermediaries that facilitate AI/ML service delivery. The network device receives capability information from the terminal, determines appropriate AI/ML models, and provides them to the terminal. This intermediary approach allows resources to be distributed appropriately between terminal and network based on actual capabilities.
2Speed
If AI/ML models are acquired and trained in real-time in terminals, then service responsiveness is improved, but resource consumption and processing load increase
Solution Approach 1:
The patent implements preliminary action by having network devices pre-train AI/ML models and store them for distribution. The terminal receives and stores these pre-trained models in advance, so when inference is needed, the terminal can immediately use the stored models without performing time-consuming training operations, thus achieving fast response with minimal energy consumption.
Solution Approach 2:
The patent applies local quality by having different functions performed at different locations: model training is performed remotely at network devices where substantial computational resources are available, while only lightweight inference operations are performed locally at the terminal. This division optimizes both responsiveness and energy efficiency.
3Productivity
If network devices allocate AI/ML tasks based on terminal capabilities, then resource utilization efficiency is improved, but information exchange complexity increases
Solution Approach 1:
The patent uses parameter changes by having the terminal report its capability parameters (processing power, storage capacity, battery status) to the network device. The network device then uses these parameters to determine which AI/ML models and tasks are appropriate for the terminal. This parameter-based approach enables efficient resource allocation through standardized information exchange.
Data Source
Figure 1~4
Figure 5
Figure 6
AI summary
The present application relates to an information reporting method, apparatus and device, and a storage medium. The method comprises: a terminal sends AI/ML capability information to a network device, the AI/ML capability information indicating resource information of a terminal for processing an AI/ML service; the network device, according to the AI/ML capability information reported by the terminal, can flexibly switch an AI/ML model run by the terminal, distribute an appropriate AI/ML model for the terminal, and adjust AI/ML training parameters and the like. Therefore, while it is ensured that an AI/ML task can be completed, AI/ML resources such as the processing capability, storage capability, and battery of the terminal can be utilized more efficiently, so that the reliability, timeliness and efficiency of AI/ML operations based on a terminal are ensured.