CSI Compression Model Signaling for Lower Decoder Complexity
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing technologies face challenges in efficiently compressing and decompressing channel state information (CSI) across communication devices, leading to increased burdens in training and maintaining multiple models at both terminal devices and network devices.
Innovation Solution
Implementing a terminal device-first training scheme where a common decompression model is used across multiple compression models, aided by an indicator corresponding to each compression model, and similarly, a network device-first training scheme where a common compression model is used across multiple decompression models, aided by an indicator corresponding to each decompression model.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If multiple compression models are used at terminal devices to improve CSI compression performance, then compression accuracy is improved, but device complexity and training burden increase
Solution Approach 1:
The patent introduces a universal decompression model at the network device that can handle multiple different compression models. This single multi-functional model replaces what would otherwise require multiple specialized decompression models, reducing the number of models that need to be maintained while preserving the ability to accurately decompress CSI from various compression approaches
Solution Approach 2:
The indicator information acts as an intermediary that bridges the terminal device's compression model selection and the network device's decompression process. By transmitting this indicator, the system enables the universal decompression model to adapt to different compression models without requiring multiple decompression models, thus resolving the contradiction between compression accuracy and system complexity
2Measurement precision
If multiple decompression models are maintained at network devices to support different compression models, then decompression accuracy is improved, but training and maintenance costs increase
Solution Approach 1:
The patent employs a universal decompression model at the network device that serves multiple functions by accommodating different compression models. This single model with multi-functional capability eliminates the need to train and maintain multiple separate decompression models, directly addressing the contradiction between decompression accuracy and the quantity of models required
Solution Approach 2:
The system uses parameter changes in the form of indicator information that specifies which compression model was used. This parameter allows the universal decompression model to adjust its processing accordingly, maintaining high decompression accuracy across different compression models without requiring multiple specialized models
3Productivity
If AI/ML models are used for CSI compression to improve communication quality, then communication performance is improved, but training burden and computational resources increase
Solution Approach 1:
By creating a universal decompression model that can handle multiple compression models, the system reduces the overall training burden. Instead of training separate decompression models for each compression approach, a single universal model is trained, significantly reducing the time and computational resources required while maintaining improved communication quality through AI/ML-based CSI compression
Data Source
AI summary
A method comprising: generating, at a terminal device, channel state information for a channel between the terminal device and a network device; generate compressed channel state information based on the channel state information by using a compression model; and transmit, to the network device, the compressed channel state information and an identification of the compression model.