Precoding Matrix Indicator Generation via Orthogonal Matching Pursuit
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current 3GPP standards face challenges in accurately compressing channel state information (CSI) for FDD systems, leading to information loss and inefficiencies in beamforming and multi-user multiplexing.
Innovation Solution
The method employs orthogonal matching pursuit (OMP) processing to generate a precoding matrix indicator (PMI) from a codebook of precoding matrices, reducing CSI compression error and improving processing efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If type I codebook with 2D DFT operation is used, then the system can provide basic CSI compression, but the manufacturing precision of CSI representation deteriorates as more antennas are deployed
Solution Approach 1:
The patent transitions from type I codebook with basic 2D DFT operation to type II codebook with enhanced parameters including increased oversample factors (O1, O2), introduction of beam clustering with L beams, and linear combination coefficients (cr,l,i). These parameter changes enable the codebook to accurately represent CSI for systems with more antennas while maintaining manageable complexity through structured beam selection and coefficient quantization.
2Manufacturing precision
If type II codebook with linear combination of multiple beams is used, then the manufacturing precision of CSI representation is improved, but the device complexity increases
Solution Approach 1:
The patent segments the complex type II codebook structure into manageable components: beam selection (identifying L dominant beams from the codebook), beam clustering (grouping beams by spatial characteristics), and coefficient determination (calculating linear combination weights). This segmentation allows the system to achieve high CSI representation accuracy through structured processing steps rather than brute-force complexity.
Solution Approach 2:
The patent introduces additional dimensions to the codebook structure beyond simple 2D DFT, including the beam combination dimension (L beams per cluster), polarization dimension (dual-polarisation handling), and frequency dimension (wideband and subband coefficients). These dimensional extensions enable accurate CSI representation for massive MIMO while organizing complexity across multiple structured axes.
3Manufacturing precision
If higher beam count L is used in type II codebook, then the manufacturing precision is improved, but the loss of information in feedback resources increases
Solution Approach 1:
The patent applies partial action by selecting only L dominant beams from the full codebook set of 4N1O1×N2O2 beam vectors. Instead of utilizing all available beams, the system identifies and processes only the most significant ones through beam selection algorithms, achieving high spatial resolution with reduced feedback overhead by focusing computational resources on the most impactful beam components.
4Manufacturing precision
If fine quantization of amplitude and phase coefficients is used, then the manufacturing precision is improved, but the quantity of feedback information increases
Solution Approach 1:
The patent implements dynamic quantization where the precision of amplitude and phase coefficients is adapted based on channel conditions and configuration parameters. The system allows flexible adjustment of quantization bits for wideband amplitude (pr,l,i(1)) and subband amplitude variations (pr,l,i(2)), enabling the feedback overhead to be optimized according to the actual spatial resolution requirements and channel variability rather than using fixed high precision for all coefficients.
Data Source
AI summary
A method of precoding matrix indicator, PMI, generation for channel state information, CSI, compression in a communication system comprising a first radio transceiver device, RTD, and a second RTD, the method comprising: transmitting, from the second RTD to the first RTD, a reference signal; receiving, at the first RTD from the second RTD, the reference signal; estimating, at the first RTD, CSI based on the received reference signal; generating, at the first RTD, a PMI to perform CSI compression, wherein the PMI indicates a precoding matrix, selected from among a codebook of precoding matrices, based on orthogonal matching pursuit, OMP, processing of the estimated CSI; generating, at the first RTD, a compressed CSI based on the generated PMI for transmission to the second RTD; and transmitting the compressed CSI from the first RTD to the second RTD.