AI/ML CSI Compression Model Performance Evaluation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current AI/ML-based CSI compression methods require significant bandwidth for performance monitoring, which is inefficient in wireless communication networks.
Innovation Solution
A method is proposed where a User Equipment (UE) compresses CSI using an encoder and transmits it to a Base Station (BS), which reconstructs the CSI using a decoder. The performance of the AI/ML model is evaluated by comparing Channel Quality Indicators (CQIs) at both the UE and BS, minimizing bandwidth usage.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If ground truth CSI is transferred to the base station for performance monitoring, then the AI/ML model performance can be evaluated, but the bandwidth consumption increases significantly
Solution Approach 1:
The patent extracts only the essential performance metric (CQI) from the complete CSI data for transmission to the base station, rather than transferring the entire ground truth CSI matrix. This selective extraction maintains performance monitoring capability while significantly reducing the overhead bandwidth consumption.
Solution Approach 2:
Instead of transferring CSI from receiver to transmitter as in conventional approaches, the patent inverts the approach by having the transmitter generate artificial ground truth CSI locally using the reconstructed channel and codebook, eliminating the need for large-scale CSI transfer while enabling performance monitoring.
2Measurement precision
If reconstructed CSI is transferred to the UE for performance monitoring, then the AI/ML model performance can be evaluated, but the bandwidth consumption increases significantly
Solution Approach 1:
The patent extracts only the CQI metric from the reconstructed CSI for transmission to the UE, rather than transferring the complete reconstructed CSI matrix. This selective extraction enables performance monitoring at the UE while minimizing the overhead bandwidth required for feedback transmission.
3Measurement precision
If complete channel feedback is transmitted, then the channel state information accuracy is improved, but the feedback overhead increases
Solution Approach 1:
The patent extracts only the essential CQI metric from the complete channel state information for feedback transmission. This selective extraction provides the transmitter with sufficient channel quality information for performance monitoring and adaptive transmission while significantly reducing the feedback overhead compared to transmitting complete CSI matrices.
Data Source
AI summary
The present invention describes a method of evaluating performance of a two-sided model used for performing CSI compression. The method includes receiving a first reference signal in a first time slot (t). A first channel Ht and a first Channel Quality Indicator (CQIt) are estimated at a first time instance. The first channel (Ht) is compressed using an encoder of a two-sided model. A compressed channel (Ht) along with the CQIt is transmitted to a Base Station (BS) (102). A second reference signal precoded with a reconstructed channel (Ĥt) is received in a second time slot (t+1). A second channel Ht+1 and a second CQIt+1 are estimated at a second time instance. The second CQI (CQIt+1) and the first CQI (CQIt) are compared for determining performance of the encoder of the two-sided model.