Deep Learning-Assisted TeraHertz QPSK Detection Relying on Single-Bit Quantization

TeraHertz (THz) wireless communication constitutes a promising technique of satisfying the ever-increasing appetite for high-rate services. However, ultra-wide bandwidth THz communications requires high-speed, high-resolution analog-to-digital converters, which are hard to implement due their high complexity and power consumption. In this paper, deep learning-assisted receiver is designed, relies on single-bit quantization. Specifically, imperfections devices, including in-phase/quadrature-phase imbalance, phase noise nonlinearity investigated. The deflection ratio maximum-likelihood detector used by our single-bit-quantization derived, reveals effect offset demodulation performance, guiding architecture design proposed receiver. To combat performance loss caused above-mentioned distortions, twin-phase training strategy neural network based demodulator proposed, where received signal compensated before sampling. Our simulation results demonstrate that capable achieving satisfactory bit error rate despite grave distortions encountered.