CSI Feedback Using Dimension Reduction Matrix and Eigenvector
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for obtaining channel state information (CSI) in Massive MIMO systems are limited by the accuracy of CSI representation using precoding matrices, leading to incomplete channel state information feedback, which affects the network device's ability to achieve optimal spatial multiplexing gains.
Innovation Solution
A channel state information feedback method that uses a dimension reduction matrix and an eigenvector to accurately represent CSI, where the terminal device sends matrix information and vector information based on downlink reference signals, allowing the network device to determine a precoding matrix for improved CSI accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a precoding matrix from a stored codebook is used to represent CSI, then the feedback mechanism is simple and standardized, but the accuracy of CSI representation is limited
Solution Approach 1:
The precoding matrix is segmented into two independent components: a dimension reduction matrix (W1) that performs spatial dimensionality reduction, and an eigenvector matrix (W2) that captures channel characteristics. This segmentation allows each component to be optimized independently, with W1 reducing complexity and W2 preserving accuracy, thereby resolving the contradiction between simplicity and precision in CSI feedback
Solution Approach 2:
The patent transitions from representing CSI using a single precoding matrix to using a two-matrix decomposition structure. By introducing the dimension reduction matrix that operates in the spatial dimension and the eigenvector matrix that operates in the channel characteristic dimension, the system achieves more accurate CSI representation while maintaining manageable feedback overhead through the structured decomposition
2Productivity
If the network device obtains accurate CSI, then spatial multiplexing gains are improved, but feedback overhead increases
Solution Approach 1:
The patent extracts and separates the essential channel characteristics into the eigenvector matrix W2, while the dimension reduction matrix W1 captures the spatial structure. By extracting only the most significant components needed for accurate CSI representation, the system reduces feedback overhead compared to transmitting complete channel state information, while still enabling accurate precoding for spatial multiplexing
Solution Approach 2:
Instead of transmitting complete channel state information, the patent uses partial action by transmitting only the decomposed matrix indices and parameters that are sufficient for accurate CSI reconstruction at the network device. This partial feedback approach achieves the necessary accuracy for spatial multiplexing while significantly reducing feedback overhead
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Embodiments of the present invention provide a channel state information feedback method, a terminal device, and a network device, so that the network device can obtain highly accurate CSI. The method includes: sending, by a terminal device, matrix information of a dimension reduction matrix to a network device, where a first dimension of the dimension reduction matrix is the same as a quantity of transmit antenna ports of the network device, and a second dimension of the dimension reduction matrix is less than the first dimension of the dimension reduction matrix; and sending, by the terminal device, vector information of an eigenvector of a downlink equivalent channel to the network device, where the eigenvector of the downlink equivalent channel is obtained based on the dimension reduction matrix, where the matrix information includes a matrix index of the dimension reduction matrix, or the matrix information includes information obtained by the terminal device by quantizing an element of the dimension reduction matrix; and the vector information includes a vector index of the eigenvector, or the vector information includes information obtained by the terminal device by quantizing an element of the eigenvector.