
Research Article
Millimeter Wave Hybrid Precoding Based on Deep Learning
@INPROCEEDINGS{10.1007/978-3-030-99200-2_2, author={Qing Liu and Ken Long}, title={Millimeter Wave Hybrid Precoding Based on Deep Learning}, proceedings={Communications and Networking. 16th EAI International Conference, ChinaCom 2021, Virtual Event, November 21-22, 2021, Proceedings}, proceedings_a={CHINACOM}, year={2022}, month={4}, keywords={Millimeter wave Deep learning Hybrid precoding MIMO}, doi={10.1007/978-3-030-99200-2_2} }
- Qing Liu
Ken Long
Year: 2022
Millimeter Wave Hybrid Precoding Based on Deep Learning
CHINACOM
Springer
DOI: 10.1007/978-3-030-99200-2_2
Abstract
To overcome the high energy consumption, and insufficient use of spatial information in traditional hybrid precoding algorithms, a hybrid precoding algorithm based on deep learning was proposed for millimeter wave massive MIMO system. Firstly, the hybrid precoding design problem was transformed into an exhaustive search problem for the analog precoder/combiner using the equivalent channel matrix. Then, a deep learning model was constructed to learn how to optimize the cascaded hybrid precoder by an improved convolutional neural network model. Finally, the optimized cascaded hybrid precoder was used to predict the analog precoding/combiner matrix directly, and the digital precoding/combiner matrix was obtained by applying singular value decomposition (SVD) to the equivalent channel matrix. Simulation results show that the performance of the proposed cascaded hybrid precoder is close to that of the pure digital precoder and can maximize the achievable rate to enhance the spectral efficiency of the millimeter wave massive MIMO system.