Channel Estimation Method Based on Transformer in High Dynamic Environment

As the relative movement speeds of the communication parties increase, the Doppler frequency offset gradually increases, and the speed of channel state information(CSI) change also increases, which limits the performance of traditional channel estimation algorithms. To solve the above problems, we p...

Full description

Saved in:
Bibliographic Details
Published in2020 International Conference on Wireless Communications and Signal Processing (WCSP) pp. 817 - 822
Main Authors Chen, Zhuolin, Gu, Fanglin, Jiang, Rui
Format Conference Proceeding
LanguageEnglish
Published IEEE 21.10.2020
Subjects
Channel estimation
Deep learning
fast time-varying channel
Feature extraction
Frequency estimation
Heuristic algorithms
OFDM
Signal processing algorithms
Online AccessGet full text

Cover

More Information
Summary:As the relative movement speeds of the communication parties increase, the Doppler frequency offset gradually increases, and the speed of channel state information(CSI) change also increases, which limits the performance of traditional channel estimation algorithms. To solve the above problems, we propose a channel estimation structure based on Transformer. Convolutional Neural Network (CNN) is used to extract the feature vectors of channel response and Transformer is used for channel estimation. Utilize Transformer's deep learning capabilities to better track channel variation characteristics in highly dynamic environments. By simulation, we get results that compared with traditional channel estimation methods, the performance of the proposed channel estimation method is improved significantly under high dynamic environment.
ISSN:2472-7628
DOI:10.1109/WCSP49889.2020.9299821