CSI Feedback Compression via Linear Combination Coefficient Selection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current channel state information (CSI) feedback mechanisms in New Radio (NR) type II face significant overhead challenges, particularly when extending beyond rank=2 transmissions, due to linear scaling of feedback overhead with the rank of CSI, which hinders efficient MU-MIMO performance and increases uplink resource requirements.
Innovation Solution
A method involving the compression of CSI by selecting a subset of linear combination coefficients from a two-dimensional matrix, excluding the lowest index column, and reporting only the locations and values of the selected coefficients, thereby reducing the feedback overhead while maintaining performance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If legacy Type II CSI feedback framework is extended to support higher-rank transmissions, then transmission rank capability is improved, but feedback overhead increases linearly with rank
Solution Approach 1:
The patent extracts and reports only the most significant linear combination coefficients from the CSI feedback data. By identifying and selecting a subset of coefficients that contribute most to channel state accuracy (based on magnitude thresholds or significance criteria), the system transmits only essential information, thereby reducing feedback overhead while maintaining adequate CSI quality for higher-rank transmissions.
Solution Approach 2:
The patent applies different reporting strategies to different components of the CSI feedback based on their importance. Significant coefficients (those above a threshold or in critical positions) are reported with full precision, while less significant coefficients are either omitted or reported with reduced precision. This differential treatment optimizes the balance between feedback accuracy and overhead for each transmission rank.
2Measurement precision
If all linear combination coefficients are reported for CSI feedback, then CSI accuracy is improved, but uplink resource requirements increase
Solution Approach 1:
The patent extracts only the significant linear combination coefficients from the complete set of CSI parameters. By applying selection criteria (such as magnitude thresholds, sparsity patterns, or importance weighting) to identify which coefficients materially contribute to channel state accuracy, the system transmits a compressed subset that maintains CSI quality while reducing uplink resource consumption for feedback transmission.
Solution Approach 2:
The patent changes the reporting parameters by introducing significance thresholds and selection criteria that determine which coefficients are reported. Instead of uniformly reporting all coefficients, the system dynamically adjusts which parameters are transmitted based on channel conditions, rank, and significance metrics, thereby optimizing the trade-off between CSI accuracy and uplink resource usage.
3Quantity of substance
If feedback overhead is reduced through coefficient selection, then uplink resources are saved, but CSI feedback completeness may be compromised
Solution Approach 1:
The patent extracts significant coefficients that capture the essential channel characteristics. By using selection criteria based on coefficient magnitude, spatial significance, or correlation metrics, the system identifies and reports only those coefficients that contain meaningful channel information, thereby reducing feedback overhead while preserving the completeness of critical CSI data needed for accurate channel estimation and precoding.
Solution Approach 2:
The patent implements feedback mechanisms where the selected coefficients are used to reconstruct the channel state at the receiver, and the quality of this reconstruction is monitored. The feedback loop ensures that the subset of reported coefficients adequately represents the channel, allowing the system to adapt the selection criteria to maintain CSI completeness while minimizing feedback overhead.
Data Source
AI summary
In accordance with an example embodiment of the present invention, a method comprising: selecting, by a user equipment, a subset of linear combination coefficients from a linearized two-dimensional matrix having columns of frequency domain components and rows of spatial beams components for channel state information determination, wherein the number of linear combination coefficients in the subset is less than all of the linear combination coefficients; determining indication comprising information associated with column indices of the selected subset of linear combination coefficients from the linearized two-dimensional matrix, wherein the indication excludes the index of the column with lowest index of the linearized two-dimensional matrix; determining compressed channel state information comprising locations in the linearized two-dimensional matrix of the subset of linear combination coefficients and corresponding values of the linear combination coefficients at those locations; and reporting, from the user equipment toward the base station, the compressed channel state information.