AI/ML-Based CSI Reporting Segmentation for Wireless Signal Efficiency
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing wireless communication systems face inefficiencies in performing wireless signal transmission and reception procedures.
Innovation Solution
A method for transmitting channel state information (CSI) by user equipment (UE) in a wireless communication system, involving receiving a reference signal, calculating CSI, determining multiple CSI parts, and transmitting a CSI report. The CSI includes AI/ML model-based information, with specific details on the number of bits and accuracy for each layer.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If conventional CSI transmission methods are used, then the system maintains simplicity in implementation, but wireless signal transmission and reception procedures become inefficient
Solution Approach 1:
The CSI report is divided into multiple parts (first CSI part, second CSI part, etc.), where each part contains specific types of CSI information. This segmentation allows the system to transmit detailed CSI efficiently by organizing information into structured parts, improving transmission efficiency without overwhelming the system with undifferentiated data complexity.
Solution Approach 2:
The patent introduces AI/ML model-based CSI as a new dimension of information representation. Instead of using traditional CSI formats, the system employs AI/ML models to generate and transmit CSI, enabling more efficient wireless signal transmission by leveraging the pattern recognition and prediction capabilities of AI/ML algorithms.
2Measurement precision
If detailed CSI information is transmitted for each layer, then communication performance improves, but the amount of data to be transmitted increases
Solution Approach 1:
The system transmits CSI for each layer only when necessary, using AI/ML models to determine which layers require detailed CSI reporting. This partial action approach maintains high measurement precision for critical layers while reducing the overall quantity of CSI data transmitted by omitting redundant information from layers that do not require detailed reporting.
Solution Approach 2:
The patent changes the parameter representation of CSI by using AI/ML model outputs instead of traditional CSI parameters. This transformation allows the system to convey accurate channel state information in a compressed format, improving measurement precision while reducing the quantity of data that needs to be transmitted and processed.
3Productivity
If AI/ML model-based CSI is implemented, then wireless transmission efficiency improves, but the complexity of processing and transmitting model information increases
Solution Approach 1:
The AI/ML models are trained and configured in advance before actual CSI transmission occurs. This preliminary action allows the models to be optimized and ready for deployment, improving wireless transmission efficiency during operation while minimizing the processing complexity during real-time CSI generation by using pre-computed model parameters.
Solution Approach 2:
The patent introduces an intermediary layer between traditional CSI generation and transmission by using AI/ML models as mediators. These models process raw channel information and transform it into optimized CSI representations, improving transmission efficiency while managing processing complexity by offloading complex computations to the AI/ML intermediary layer.
Data Source
Figure 1~2
Figure 3
Figure 4~5
AI summary
A terminal according to at least one of embodiments disclosed in the present specification may: receive a reference signal for CSI from a base station; calculate CSI on the basis of the reference signal; determine a plurality of CSI parts including a first CSI part and a second CSI part on the basis of the CSI; and transmit a CSI report on the basis of the plurality of CSI parts, wherein the calculated CSI includes artificial intelligence/machine learning (AI/ML) model-based CSI, the first CSI part includes information on a total number of bits of the AI/ML model-based CSI for all layers, and the second CSI part includes information on the number of bits for each layer of the AI/ML model-based CSI.