Deep Learning Based Pilot-Free Transmission: Error Correction Coding for Low-Resolution Reception Under Time-Varying Channels

Recently, deep learning aided methods have been developed for error correction coding with quantitative constraints. However, previous studies only focus on additive white Gaussian noise (AWGN) channels, which is not sufficient for actual communication environments. In this paper, we propose a novel autoencoder aided error correction coding scheme for low-resolution reception under time-varying channels. Based on the symbol extension of the proposed autoencoder and the faster-than-Nyquist (FTN) technology, pilot-free transmission can be realized without adding additional bandwidth. The transformer block is introduced to lighten and improve the decoder. Additionally, two kinds of preamplification techniques are applied for further performance boosting. Simulations show that the proposed method can achieve better performance compared with the traditional methods at high signal-to-noise ratio (SNR) under different time-varying channels without quantization. Moreover, it outperforms the previous state-of-the-art ECCNet and can achieve remarkable transmission performance even under time-varying low-resolution reception scenarios.