DL-Based OTFS Signal Detection in Presence of Hardware Impairments

Abstract

Orthogonal time frequency space (OTFS) modulation is an emerging technique for next-generation communication due to its robustness to the doubly dispersive channels under high mobility scenarios. We have designed and analyzed a deep learning (DL)-based OTFS system (DL-OTFS) in the presence of hardware impairments (HI) such as in-phase and quadrature-phase (IQ) component mismatch and DC offset. Further, data augmentation is also considered for the proposed DL-OTFS to enhance the system performance. Numerical results show that the DL-OTFS model can efficiently learn the input and output relation and leads to improved bit error rate (BER) performance than the conventional message passing and minimum mean square error (MMSE)-based receiver with and without HI.