Analysis of an LSTM-based NOMA Detector Over Time Selective Nakagami-m Fading Channel Conditions

Abstract

This work examines the efficacy of deep learning (DL) based non-orthogonal multiple access (NOMA) receivers in vehicular communications (VC). Analytical formulations for the outage probability (OP), symbol error rate (SER), and ergodic sum rate for the researched vehicle networks are established using i.i.d. Nakagami-m fading links. Standard receivers, such as least square (LS) and minimum mean square error (MMSE), are outperformed by the stacked long-short term memory (S-LSTM) based DL-NOMA receiver. Under real time propagation circumstances, including the cyclic prefix (CP) and clipping distortion, the simulation curves compare the performance of MMSE and LS receivers with that of the DL-NOMA receiver. According to numerical statistics, NOMA outperforms conventional orthogonal multiple access (OMA) by roughly 20% and has a high sum rate when considering i.i.d. fading links

Keywords

• deep learning (DL)
• multiple-input multiple-output (MIMO)
• non orthogonal multiple access (NOMA)
• orthogonal multiple access (OMA)