Base Station Channel Super-Resolution for Low-Overhead PMI
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current communication systems face challenges in accurately reconstructing channels due to large quantization granularity in PMI information, leading to increased quantization noise and reduced effectiveness of MIMO precoding in eliminating inter-user interference.
Innovation Solution
A base station employs a super-resolution network to perform interpolation and denoising on channel information of a first granularity, obtaining a second channel with finer granularity for improved downlink precoding, and uses neural networks for channel reconstruction and feedback mechanisms to enhance channel estimation accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If subband-level PMI information is used for channel reconstruction, then feedback overhead is reduced, but quantization noise increases and channel estimation accuracy deteriorates
Solution Approach 1:
The patent segments the channel reconstruction process into two stages: first reconstructing subband-level channels using compressed PMI information, then generating full-bandwidth channels by combining these subband channels with channel difference information. This segmentation allows efficient feedback while maintaining accuracy through the two-stage reconstruction approach.
Solution Approach 2:
The patent introduces channel difference information as an intermediary element that bridges the gap between compressed subband-level PMI information and full-resolution channels. This intermediary carries the fine-grained channel variations that are lost in the compression process, enabling accurate channel estimation without increasing PMI feedback overhead.
2Device complexity
If subband-level PMI information is used, then feedback complexity is reduced, but quantization noise in space domain and coefficients increases
Solution Approach 1:
The patent performs preliminary channel reconstruction at subband level using PMI information, then applies a second processing stage using channel difference information to eliminate quantization noise. This preliminary action approach allows the system to first establish a baseline reconstruction and then refine it to remove harmful quantization effects.
Solution Approach 2:
The patent maintains continuous useful action by combining the subband-level channel reconstruction with the channel difference information in a unified two-stage process. This continuity ensures that the beneficial low-complexity PMI feedback is continuously enhanced by the channel difference information, systematically reducing quantization noise throughout the reconstruction process.
3Adaptability or versatility
If PRB bundling granularity is smaller than PMI granularity, then resource allocation flexibility is improved, but channel reconstruction accuracy deteriorates due to mismatched granularities
Solution Approach 1:
The patent applies local quality by processing different frequency subbands with appropriate granularity. The subband-level PMI provides coarse-grained information for broad frequency ranges, while the channel difference information provides fine-grained adjustments for specific PRBs. This local differentiation allows flexible resource allocation at PRB level while maintaining accurate channel reconstruction through the combination of coarse and fine information.
Data Source
AI summary
The present disclosure provides a base station and a terminal. The base station includes: a receiving unit configured to receive precoding matrix indicator information of a first granularity from a terminal; and a processing unit configured to perform channel reconstruction according to the precoding matrix indicator information to obtain a first channel, use a super-resolution network to perform interpolation and denoising processing on a channel having the first granularity to obtain a second channel, and perform downlink precoding on the second channel, wherein the first channel has the first granularity, the second channel has a second granularity, and the second granularity is finer than the first granularity.