Autoencoder for end-to-end learning communication system based on NOMA

For the future wireless communication system, non-orthogonal multiple access (NOMA) has attracted much attention as a key candidate in recent years. The advent of Deep Learning (DL) into the realm of wireless communication has created a new paradigm in the design of the physical layer. Deep learning based Autoencoder (AE) is a promising tool to create a full-fledged communication system. In the field of neural network-based communication systems, there has been a lot of research done, to find near optimal solutions in the past few years. In this paper, the combination of both AE and NOMA for end-to-end transmission is proposed. The proposed AE-NOMA signal detection method shows feasible improvement of 6 dB performance gain for user1 and user2 in terms of signal-to-noise ratio (SNR) as compared to previous Deep neural network-based NOMA systems.