Configured Reference Signal Selection for AI/ML Beam and CSI Feedback
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing communication technologies face challenges in efficiently managing artificial intelligence/machine learning (AI/ML) models for beam management and channel state information (CSI) feedback due to high overhead in reporting data requirements, particularly in scenarios like 5G millimeter wave communication, where beamforming and CSI compression are crucial.
Innovation Solution
A method and system where a network device transmits configuration information to a terminal device, including reference signal (RS) resources, enabling the terminal device to determine and manage an AI/ML model based on these RS resources, thereby reducing overhead by optimizing the selection and use of training RSs for beam management and CSI feedback.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of energy
If traditional methods are used for beam management and CSI feedback in 5G millimeter wave communication, then communication coverage and signal quality can be maintained, but data reporting overhead becomes excessively high
Solution Approach 1:
The patent extracts only the essential information needed for beam management and CSI feedback by using AI/ML models to process and compress channel state information. Instead of reporting all raw channel data, the system extracts key features and parameters that are most relevant for maintaining communication quality, thereby reducing overhead while preserving reliability.
Solution Approach 2:
The patent changes the parameter representation from traditional detailed channel state information to compressed AI/ML model parameters. By transforming the data format and using neural network models to represent channel conditions, the system achieves more efficient data transmission with reduced overhead while maintaining the necessary information for reliable beam management.
2Measurement precision
If detailed channel state information is reported for accurate beam management, then beamforming performance is improved, but the complexity and overhead of data transmission increases
Solution Approach 1:
The patent uses AI/ML models to create compressed representations or 'copies' of the channel state information. Instead of transmitting the full detailed channel data, the system transmits parameters that can be used to reconstruct or approximate the channel state at the receiver, achieving a balance between measurement precision and transmission complexity.
3Reliability
If traditional reference signal resources are used for AI/ML model training, then model accuracy can be maintained, but the overhead for configuring and transmitting training data increases
Solution Approach 1:
The patent performs preliminary configuration of reference signal resources by the network device, preparing training data in advance before AI/ML model training is needed. This preliminary action allows the terminal device to receive pre-configured reference signal resources and use them directly for model training without requiring extensive real-time configuration exchanges, thereby reducing configuration overhead while maintaining model accuracy.
Data Source
AI summary
Embodiments of the present disclosure relate to methods, devices, and computer readable medium for communication. According to embodiments of the present disclosure, a network device transmits configuration information to a terminal device. The configuration information is associated with a determination of at least one set of RS resources. The terminal device determines a set of training reference signals for management of an AI/ML model based on the at least one set of RS resources. The terminal device updates the AI/ML model based on the set of training reference signals. In this way, the overhead can be reduced.