CSI Feedback Compression via Eigenvalue Decomposition

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current 5G communication systems face challenges in efficiently acquiring and compressing channel state information (CSI) feedback, particularly with the increasing number of antenna ports, which affects beamforming accuracy and system throughput.

Innovation Solution

The proposed solution involves a user equipment (UE) that receives CSI feedback configuration information and determines a CSI matrix, identifying and transmitting a set of basis vectors and coefficients using principal component analysis (PCA) for dimensionality reduction, allowing for efficient CSI feedback over an uplink channel, enabling accurate beamforming and reduced overhead.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Power

If the number of antenna ports is increased to support massive MIMO and beamforming in 5G systems, then beamforming gain and system capacity are improved, but the dimensionality and overhead of CSI feedback data increases significantly

Engineering Contradiction:
Improvebeamforming gainVSAvoidCSI feedback data volume
Core Design Contradiction:
PowerVSQuantity of substance

Solution Approach 1:

The patent extracts only the essential channel characteristics by performing eigenvalue decomposition and selecting only the dominant eigenvectors corresponding to the largest eigenvalues. This extraction process separates the most significant channel information from the complete channel state data, transmitting only the necessary components (eigenvectors and eigenvalues) rather than the full CSI matrix, thereby reducing feedback overhead while preserving beamforming performance

Inventive Principle:
Principle #2Taking out (Extraction)

Solution Approach 2:

The patent creates a compressed representation (copy) of the channel state information using eigenvalue decomposition. Instead of transmitting the complete channel matrix, the system transmits a simplified copy consisting of eigenvectors and eigenvalues that capture the essential channel characteristics. This compressed copy maintains sufficient accuracy for beamforming operations while dramatically reducing the amount of data that needs to be fed back to the base station

Inventive Principle:
Principle #26Copying

2Measurement precision

If the complete CSI matrix is transmitted for accurate channel representation, then beamforming accuracy is maintained, but the feedback overhead and processing complexity increase

Engineering Contradiction:
Improvechannel representation accuracyVSAvoidfeedback processing complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent extracts only the essential channel characteristics by performing eigenvalue decomposition and selecting only the dominant eigenvectors corresponding to the largest eigenvalues. This extraction process separates the most significant channel information from the complete channel state data, transmitting only the necessary components (eigenvectors and eigenvalues) rather than the full CSI matrix, thereby reducing feedback overhead while preserving beamforming performance

Inventive Principle:
Principle #2Taking out (Extraction)

Solution Approach 2:

The patent transforms the channel state information from the original channel matrix domain to the eigenvalue decomposition domain. By changing the representation parameters from complete channel coefficients to eigenvalues and eigenvectors, the system achieves a more compact and efficient representation. This parameter transformation allows the essential channel characteristics to be captured with fewer parameters, reducing both feedback overhead and processing complexity

Inventive Principle:
Principle #35Parameter changes

3Quantity of substance

If dimensionality reduction techniques are applied to compress CSI feedback, then feedback overhead is reduced, but potential loss of channel information may occur

Engineering Contradiction:
Improvefeedback data volumeVSAvoidchannel information loss
Core Design Contradiction:
Quantity of substanceVSLoss of information

Solution Approach 1:

The patent transforms the channel state information from the original channel matrix domain to the eigenvalue decomposition domain. By changing the representation parameters from complete channel coefficients to eigenvalues and eigenvectors, the system achieves a more compact and efficient representation. This parameter transformation allows the essential channel characteristics to be captured with fewer parameters, reducing both feedback overhead and processing complexity

Inventive Principle:
Principle #35Parameter changes

Solution Approach 2:

The patent applies partial action by transmitting only the dominant eigenvectors and eigenvalues that capture the most significant channel characteristics, rather than transmitting complete CSI or all eigenvectors. The system determines an appropriate number of dominant components based on channel conditions, transmitting sufficient information to maintain beamforming accuracy while minimizing feedback overhead. This partial transmission strategy prevents information loss of the most critical channel aspects

Inventive Principle:
Principle #16Partial or excessive action

Data Source

PatentEP3510707B1Method and apparatus to enable channel compression in advanced wireless communication systems
Publication Date: 2020.12.30 SAMSUNG ELECTRONICS CO LTD
  • EP3510707B1 patent drawingFigure 1
  • EP3510707B1 patent drawingFigure 2
  • EP3510707B1 patent drawingFigure 3

AI summary

A method of a user equipment (UE) for a channel state information (CSI) feedback in a wireless communication system. The method comprises receiving, from a base station (BS), CSI feedback configuration information for the CSI feedback including a spatial channel information (SCI) indicator for each subband (SB), wherein the SCI indicator indicates a SCI associated with the downlink (DL) channel matrix; determining a CSI matrix H K,N comprising a dimension K x N based on the CSI feedback configuration information, where K indicates a number of SBs and N indicates a number of components of the SCI; identifying, based on the CSI matrix H K,N , the SCI indicator that indicates a first set of d basis vectors comprising a dimension K x 1, a second set of d basis vectors comprising a dimension N x 1, and a set of d coefficients, and transmitting, to the BS, the CSI feedback including the identified SCI indicator indicating the first set of d basis vectors, the second set of d basis vectors, and a set of d coefficients over an uplink channel.