CSI Feedback Codebooks From Clustered Channel Samples
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional CSI feedback methods, including codebook-based solutions and AI-based approaches, fail to provide high-performance CSI feedback in specific scenarios like macro-cell, indoor, and high-speed moving scenarios due to lack of generalization and inexplicability, hindering standardization.
Innovation Solution
Cluster CSI samples to obtain cluster centers, which serve as a CSI feedback codebook, enabling data-driven performance improvements in specific scenarios.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional codebook-based CSI feedback methods are used, then the feedback process is simple and standardized, but the CSI feedback performance is insufficient in specific scenarios like macro-cell, indoor, and high-speed moving scenarios
Solution Approach 1:
The patent applies preliminary action by pre-clustering CSI samples from historical data to generate scenario-specific codebooks before actual CSI feedback occurs. The network device stores multiple codebooks corresponding to different scenarios (macro-cell, indoor, high-speed moving) in advance, so when a specific scenario is detected, the appropriate pre-prepared codebook can be immediately used, avoiding the need for real-time adaptation and ensuring high performance in specific scenarios while maintaining standardized feedback processes
2Reliability
If AI-based CSI feedback approaches are used, then performance improvements can be achieved in specific scenarios, but the methods lack generalization and are difficult to standardize
Solution Approach 1:
The patent applies copying by creating simplified representations (codebooks) of complex channel conditions from historical data. Instead of using complex AI models for real-time processing, the system copies essential channel characteristics into pre-computed codebooks that can be easily stored, transmitted, and used in standardized feedback procedures. This allows AI-based performance improvements to be captured in a form that is easy to standardize and deploy
Solution Approach 2:
The patent applies parameter changes by transforming continuous channel state information into discrete codebook indices through clustering. The clustering process changes the parameter representation from continuous complex channel matrices to discrete codebook selections, making the system both more efficient and easier to standardize while maintaining the performance benefits of data-driven approaches
3Reliability
If cluster centers obtained by clustering CSI samples are used as CSI feedback codebook, then data-driven performance advantages are achieved in specific scenarios, but the feedback overhead increases
Solution Approach 1:
The patent applies taking out (extraction) by extracting only the essential characteristics of channel states into compact codebook representations. Instead of feeding back complete channel matrices or complex AI model outputs, the system extracts the most important information into pre-computed cluster centers that form the codebook. The terminal device only needs to feedback compact codebook indices rather than full channel information, significantly reducing feedback overhead while maintaining data-driven performance advantages
Data Source
Figure 1~2
Figure 3
Figure 4
AI summary
A method for channel state information (CSI) feedback, a transmitting device, and a receiving device are provided in implementations of the present disclosure. A set of CSI samples is clustered to obtain cluster centers, the cluster centers obtained through clustering are taken as a CSI feedback codebook, and a CSI feedback implemented based on the CSI feedback codebook can have performance advantages of specific data-driven scenarios, thereby improving the CSI feedback performance. The method for CSI feedback includes the following. A transmitting device encodes target CSI information according to the CSI feedback codebook, to obtain a target feedback bitstream. The transmitting device transmits the target feedback bitstream. The CSI feedback codebook is K cluster centers obtained by clustering CSI samples in a first dataset, the first dataset includes S CSI samples, K and S are positive integers, K=2B<S, and B is the number of CSI feedback bits.