Precoding Matrix Feedback Compression via Space-Frequency Vector Segmentation

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current precoding matrix feedback methods in massive MIMO systems result in high overheads, particularly with increasing numbers of transmission layers, which affects spectrum resource utilization and data transmission quality.

Innovation Solution

The method represents column vectors in ideal precoding matrices of each subband using linear combinations of beam vectors in the space domain and describes the change rules of weighting coefficients using frequency domain vectors, allowing for a reduced number of frequency domain vectors to describe the change rules across subbands, thereby compressing feedback overheads while maintaining approximation precision.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If wideband feedback and subband feedback are performed based on each transmission layer, then feedback precision is improved, but feedback overheads increase significantly

Engineering Contradiction:
Improvefeedback precisionVSAvoidfeedback overheads
Core Design Contradiction:
Measurement precisionVSQuantity of substance

Solution Approach 1:

The feedback information is segmented into two parts: beam selection information (wideband) and combinatorial coefficient information (subband). This segmentation allows different levels of detail to be provided at appropriate granularities, reducing overall feedback overhead while maintaining precision where needed.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent introduces a two-dimensional feedback structure combining wideband (frequency dimension) and subband (frequency detail dimension) feedback. By organizing feedback information across these dimensions, the system achieves comprehensive channel state information with reduced overhead compared to per-layer feedback.

Inventive Principle:
Principle #17Another dimension (Dimensionality change)

2Productivity

If the number of transmission layers is increased to improve spectrum resource utilization, then data transmission capability is improved, but feedback overheads increase

Engineering Contradiction:
Improvespectrum resource utilizationVSAvoidfeedback overheads
Core Design Contradiction:
ProductivityVSQuantity of substance

Solution Approach 1:

The feedback mechanism is designed to be universal across multiple transmission layers by using a common codebook structure and feedback format. The same feedback procedure can be applied regardless of the number of layers, allowing the system to scale spectrum resource utilization without proportionally increasing feedback overheads.

Inventive Principle:
Principle #6Universality (Multi-functionality)

Data Source

PatentEP3826199B1Methods for indicating and determining precoding matrix, and communication apparatus
Publication Date: 2023.10.18 HUAWEI TECH CO LTD
  • EP3826199B1 patent drawingFigure 1
  • EP3826199B1 patent drawingFigure 2
  • EP3826199B1 patent drawingFigure 3~4

AI summary

This application provides a precoding matrix indicating and determining method, and a communications apparatus, to reduce feedback overheads. The method includes: The terminal device determines a PMI, and sends the PMI to a network device. The PMI includes R groups of space-frequency information corresponding to R transmission layers, and the R groups of space-frequency information are used to determine a precoding matrix of each subband. An rth group of space-frequency information in the R groups of space-frequency information is used to indicate P x L space domain vectors corresponding to an rth transmission layer, K frequency domain vectors corresponding to the rth transmission layer, and P x L x K coefficients corresponding to the rth transmission layer. The P x L space domain vectors are determined by L beam vectors in each of P polarization directions and wideband amplitude coefficients of the L beam vectors, and each of the P x L x K coefficients is used to indicate a linear superposition coefficient of a vector operation result of one of the P x L space domain vectors and one of the K frequency domain vectors.