CSI Feedback Evaluation Using Encoder Copying at the Network Side

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing methods for evaluating CSI feedback performance in FDD mode face challenges in direct comparison of CSI at the base station and UE sides, leading to high transmission overhead and computational complexity, with AI-based methods lacking generalization and interpretability.

Innovation Solution

A method involving a network device processing CSI feedback information using an encoder and decoder to directly evaluate performance without additional transmission, utilizing metrics like bit error rate and cosine similarity to assess CSI feedback accuracy.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If direct comparison of CSI at base station and UE sides is performed, then measurement precision is improved, but transmission overhead increases

Engineering Contradiction:
ImproveCSI feedback performance evaluation accuracyVSAvoidtransmission overhead
Core Design Contradiction:
Measurement precisionVSLoss of information

Solution Approach 1:

The patent creates a virtual copy of the UE-side CSI processing pipeline at the base station by deploying the same encoder model. This allows the base station to generate first CSI feedback information from first CSI in the same way the UE generates second CSI feedback information from second CSI, enabling direct comparison without transmitting actual CSI data between devices.

Inventive Principle:
Principle #26Copying

2Measurement precision

If AI-based CSI feedback evaluation is implemented, then measurement precision is improved, but device complexity increases

Engineering Contradiction:
ImproveCSI feedback performance evaluation accuracyVSAvoidcomputational complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The base station performs self-evaluation of CSI feedback performance by using its own computational resources to run the encoder model and generate first CSI feedback information. This eliminates the need for complex coordination with the UE and reduces overall system complexity, as the base station serves itself in the evaluation process.

Inventive Principle:
Principle #25Self-service

Solution Approach 2:

The encoder model is pre-trained and deployed at the base station before evaluation is needed. This preliminary preparation allows the base station to quickly perform evaluations without requiring complex real-time computations during actual CSI feedback operations, reducing operational complexity.

Inventive Principle:
Principle #10Preliminary action

3Measurement precision

If CSI is transmitted for performance evaluation, then measurement precision is improved, but loss of information increases

Engineering Contradiction:
ImproveCSI feedback performance evaluation accuracyVSAvoidtransmission overhead
Core Design Contradiction:
Measurement precisionVSLoss of information

Solution Approach 1:

Instead of transmitting actual CSI data from the UE to the base station for evaluation, the patent creates a virtual copy of the evaluation process at the base station. The base station generates first CSI feedback information using the same encoder model that the UE uses, allowing direct comparison with second CSI feedback information without any additional CSI transmission.

Inventive Principle:
Principle #26Copying

Data Source

PatentEP4716106A1Method for evaluating CSI feedback performance, network device, and terminal device
Publication Date: 2026.03.25 GUANGDONG OPPO MOBILE TELECOMMUNICATIONS CORP LTD
  • EP4716106A1 patent drawingFigure 1~4
  • EP4716106A1 patent drawingFigure 5
  • EP4716106A1 patent drawingFigure 6

AI summary

The present disclosure relates to a method for evaluating channel state information (CSI) feedback performance, a network device, a terminal device, a chip, a computer-readable storage medium, a computer program product, a computer program, and a communication system. The network device processes a first CSI based on a first encoder to obtain a first CSI feedback information. The first CSI is obtained by the network device processing a second CSI feedback information from the terminal device based on a first decoder. The second CSI feedback information is obtained by the terminal device processing a second CSI based on the first encoder, and the second CSI is obtained based on reference signal measurement. The network device evaluates CSI feedback performance based on the first CSI feedback information. The present disclosure can accurately evaluate the CSI feedback performance without introducing significant transmission overhead.