AI Channel Compression by Layer Importance for CSI Reporting
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In high-rank scenarios, existing AI-based channel compression methods result in insufficient accuracy for important layers and excessive length for unimportant layers due to uniform compression of precoding matrices, leading to inefficient channel information reporting.
Innovation Solution
The method involves determining payload information for each layer and using a target AI unit to process channel information, ensuring the payload of the channel characteristic information matches the output length of the AI unit, thereby optimizing the compression and reporting of channel information based on layer importance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If a same encoding AI model is used to compress precoding matrices of different layers, then the compression process is simplified, but the accuracy of compression results for important layers is insufficient and the length of compression results for unimportant layers is excessively long
Solution Approach 1:
The patent applies different encoding AI models to different layers based on their importance. Specifically, a first encoding AI model is used for a first layer and a second encoding AI model is used for a second layer, allowing each layer to be compressed with appropriate quality and accuracy requirements rather than using a uniform approach across all layers
Solution Approach 2:
The patent segments the compression process by dividing layers into different groups (first layer and second layer) and applying different compression strategies to each segment. This segmentation allows important layers to receive more sophisticated compression treatment while unimportant layers use simpler compression
2Ease of operation
If uniform compression is applied to all layers, then the processing is straightforward, but the payload length does not match the output length of the AI unit requiring additional processing
Solution Approach 1:
The patent changes the parameters of the encoding AI models based on layer characteristics. Different encoding AI models with different output lengths are selected for different layers, allowing the payload information length to directly match the output length of the corresponding AI unit without requiring additional processing or padding
3Measurement precision
If different encoding AI models are used for different layers, then compression accuracy for important layers is improved, but the device complexity increases
Solution Approach 1:
The patent implements local quality by applying higher quality compression (first encoding AI model) to important layers and standard quality compression (second encoding AI model) to unimportant layers. This selective approach improves overall accuracy where needed while avoiding unnecessary complexity in areas where high accuracy is not critical
Data Source
AI summary
This application discloses an information processing method, an information transmission method, an apparatus, a terminal, and a network-side device. The information processing method in embodiments of this application includes: The terminal obtains first information, where the first information indicates payload information of first channel characteristic information corresponding to at least one layer; and the terminal performs, based on the first information, first processing on channel information of a target layer by using a target first AI unit, to obtain the first channel characteristic information corresponding to the target layer, where the payload information of the first channel characteristic information corresponding to the target layer matches an output length of the target first AI unit, and the at least one layer includes the target layer.