Metalearning-based channel estimation method under super-large scale MIMO, terminal and medium

By adopting a meta-learning-based channel estimation method in the ultra-large-scale MIMO system, building a near-field channel model and using a dense residual network, the problems of inaccurate acquisition of channel information and high computational complexity are solved, and high-precision and low-complexity channel estimation are achieved.

CN119696961APending Publication Date: 2025-03-25CHONGQING UNIV OF POSTS & TELECOMM
0 Cites 0 Cited by

Patent Information

Application Number
CN202411618787.8
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-11-13
Publication Date
2025-03-25

AI Technical Summary

Technical Problem

In hyperscale MIMO systems, existing channel estimation methods are difficult to effectively deal with spatial non-stationarity and rapidly changing channel environments, resulting in inaccurate acquisition of channel information and high computational complexity.

Method used

Using a channel estimation method based on meta-learning, a high-precision channel estimation model based on spherical waves and an image super-resolution reconstruction model of dense residual networks is constructed, and combined with meta-learning technology, the network model parameters are adjusted to adapt to different channel environments to achieve high-precision channel estimation.

Benefits of technology

It improves the accuracy and low computing complexity of channel estimation, and can quickly adapt to the new channel environment, and is suitable for ultra-large-scale MIMO systems.

✦ Generated by Eureka AI based on patent content.
Patent Text Reader

Abstract

The invention relates to the technical field of wireless communication, and provides a meta-learning-based channel estimation method under super-large-scale MIMO, a terminal and a medium, and the method comprises the steps: constructing a near-field channel model based on spherical waves based on a wireless communication system of a plurality of single-antenna users and a super-large-scale uniform linear array base station; based on an image super-resolution reconstruction model of a dense residual network, regarding channel frequency response at a pilot frequency position as a low-resolution 2D image and taking the low-resolution 2D image as input, outputting channel frequency response in a high-resolution 2D image form through the network, and reconstructing whole channel state information; according to the method, meta-learning is introduced on the basis of channel estimation of a network, the meta-learning adjusts network model parameters under a plurality of tasks, parameter updating in a new environment is realized by adopting a small number of data samples, initialization parameters are obtained, and high-precision channel estimation in different environments is realized. The method has higher estimation precision and lower calculation complexity, and can quickly adapt to a new channel environment.
Need to check novelty before this filing date? Find Prior Art