UE AI/ML Model Reconfiguration With Pause Signaling
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing wireless communication systems face challenges in efficiently managing artificial intelligence (AI)/machine learning (ML) model lifecycle operations due to limited awareness of UE capabilities and potential delays in model activation, deactivation, or switching, leading to latency and performance issues.
Innovation Solution
A method and apparatus for managing AI/ML functionalities in wireless terminals by generating messages to indicate potential delays and updating or pausing model implementations, allowing for dynamic resource management and efficient UE capability signaling.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If AI/ML model lifecycle operations are managed in existing wireless communication systems, then model functionality can be provided, but latency and performance degradation occur due to limited awareness of UE capabilities and potential delays in model activation, deactivation, or switching
Solution Approach 1:
The network node performs preliminary actions by determining potential delays associated with AI/ML model operations before actually executing the model activation, deactivation, or switching. This allows the network to plan and coordinate model lifecycle operations in advance, reducing actual execution delays and improving timing accuracy.
Solution Approach 2:
The system implements feedback mechanisms where the network node receives capability information from the wireless terminal about supported AI/ML functionalities and uses this feedback to make informed decisions about model operations. The network can adjust its model management strategies based on the terminal's actual capabilities and current state, reducing unnecessary delays.
2Measurement precision
If the network node determines potential delays for AI/ML functionality implementation, then timing accuracy improves, but additional signaling and processing are required
Solution Approach 1:
The wireless terminal performs self-service by autonomously determining its own capability information regarding supported AI/ML functionalities and reporting this information to the network node. This self-reporting mechanism reduces the network node's processing burden while still enabling accurate delay determination through the terminal's own assessments.
3Productivity
If the network configures AI/ML functionalities based on UE capabilities, then resource optimization improves, but capability signaling and configuration management become more complex
Solution Approach 1:
The capability information signaling is segmented into distinct components, with the wireless terminal reporting specific AI/ML functionality capabilities separately. This segmentation allows the network node to process and configure each capability independently, simplifying overall configuration management while enabling precise resource optimization based on actual terminal capabilities.
Solution Approach 2:
The system uses parameter changes in capability information reporting to dynamically adjust configuration. The wireless terminal can update its capability parameters as needed, and the network node adjusts AI/ML model configurations based on these parameter changes, enabling flexible resource optimization without complex manual configuration management.
Data Source
AI summary
A wireless terminal which communicates over a radio interface with a radio access network comprises receiver circuitry and processor circuitry. The receiver circuitry is configured to receive, over the radio interface from the radio access network, at least a first message which concerns implementation of an Artificial Intelligence/Machine Learning (AI/ML) Functionality/feature for the wireless terminal. The processor circuitry is configured: to make a determination that the reception of the first message requires one of the following: an update of an existing model for the Artificial Intelligence/Machine Learning (AI/ML) Functionality/feature; a download of a new model for the Artificial Intelligence/Machine Learning (AI/ML) Functionality/feature for a particular functionality or feature; and to make a determination that a pause maybe necessary to implement the Artificial Intelligence/Machine Learning (AI/ML) Functionality/feature. In an example optional implementation, the processor circuitry may also be further configured to generate a pause notification message.