LWCNet: Lightweight Complex Neural Network for Real-Time Channel Estimation

Convolutional neural networks (CNNs) have been widely adopted in various fields and the wireless communication domain is no exception; researchers leveraged CNNs for communication tasks such as channel estimation. They showed the potential of CNNs for these applications but suggested CNNs were insufficient to be considered for real world implementation due to the limited applicability and high execution latency. In this paper, we propose a Lightweight Complex Neural Network (LWCNet) composed of Complex Number Adapter (CNA), re-parameterizable multi-path operations, and progressive learning to cover various resource block configurations. LWCNet outperforms not only conventional linear channel estimation methods as least square (LS) estimator and linear minimum mean square error (LMMSE) estimator but also deep learning based estimators. Moreover, LWCNet estimates channels with various numbers of subcarriers within 0.5 ms using a modern graphics processing unit.