UE Parameter Set Activation for Low-Overhead ML Configuration
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing 5G NR technologies face challenges in efficiently managing parameter sets for machine learning operations in wireless communication, leading to significant signal overhead and storage space utilization, while lacking flexibility in adapting to device and network conditions.
Innovation Solution
A user equipment (UE) is configured to receive and activate parameter sets based on activation conditions, downloading only relevant parameter sets or delta information, thereby reducing signaling overhead and preserving storage space.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If multiple parameter sets are managed for varying device and network conditions, then adaptability is improved, but signaling overhead and storage requirements increase
Solution Approach 1:
The parameter sets are segmented into two types: type 1 parameter sets that are configured by the network and stored at the UE, and type 2 parameter sets that are downloaded from a model repository. This segmentation allows the system to manage different parameter sets through different mechanisms, reducing overall storage requirements at the UE while maintaining adaptability through selective downloading of type 2 parameter sets based on activation conditions.
Solution Approach 2:
Type 1 parameter sets are pre-configured by the network during initial setup, and type 2 parameter sets are pre-stored in the model repository. This preliminary action eliminates the need for real-time parameter set generation and reduces signaling overhead during operation, as the UE can directly activate pre-configured or pre-downloaded parameter sets based on activation conditions.
2Adaptability or versatility
If multiple parameter sets are managed for varying device and network conditions, then adaptability is improved, but signaling overhead increases
Solution Approach 1:
The patent extracts the parameter set storage function from the network to a separate model repository. This allows the network to send only compact activation indicators rather than full parameter sets, significantly reducing signaling overhead. The UE downloads parameter sets from the repository based on activation conditions, eliminating the need for repeated network transmissions.
Solution Approach 2:
Instead of transmitting full parameter sets through signaling, the system uses compact copies or references (activation indicators) that point to parameter sets stored in the model repository. This copying mechanism reduces signaling overhead while maintaining the ability to access complete parameter sets when needed.
3Quantity of substance
If parameter sets are downloaded from a model repository, then storage space at UE is preserved, but download complexity increases
Solution Approach 1:
The UE is equipped with the capability to autonomously determine activation conditions and initiate downloads of type 2 parameter sets from the model repository based on those conditions. This self-service mechanism reduces the need for complex network-controlled download procedures, simplifying the overall system while preserving UE storage space.
Data Source
AI summary
Example implementations include a method, apparatus and computer-readable medium of wireless communication by a user equipment (UE), comprising receiving parameter set configuration information from a network entity, the parameter set configuration information corresponding to a model structure employed in a machine learning operation by the UE for the wireless communication. The implementations further include activating a parameter set in response to an activation condition, the parameter set identified within the parameter set configuration information as being associated with the activation condition.