Abstract

The sixth-generation (6G) wireless network promises unprecedented enhancements in terms of system throughput, energy efficiency, traffic capacity per area, spectral efficiency, and low latency. To meet these demands, the radio interface must demonstrate the adaptability and efficient utilization of scarce frequency resources, necessitating novel multiple access techniques and new waveforms. Furthermore, in large-bandwidth multiuser networks, intersymbol interference (ISI) and inter-user interference (IUI) represent major design challenges. In this regard, time reversal (TR) has emerged as a promising 6G waveform candidate, concentrating signal energy in both the time and space domains in multipath environments. On the other hand, non-orthogonal multiple access (NOMA) promises high spectral efficiency and enhanced connectivity, serving multiple users over the same time-frequency-code resources. In this paper, we investigate the integration of NOMA and TR to mitigate these challenges and propose, for the first time in the literature, a novel receiver architecture for downlink NOMA-based TR communications, which does not require precoding at the transmitter. In more detail, power-domain NOMA is employed by the transmitter, while TR filtering is carried out at each receiving end. We derive novel approximated expressions for the pairwise error probability (PEP), a fundamental component in establishing the union bound on the bit error rate (BER), to characterize users' performance. We extensively employ Monte Carlo simulations and numerical analyses to verify the analytical expressions, providing significant insights into the error rate performance for each user. Also, we investigate the performance gain of the proposed NOMA-based TR receiver, over the orthogonal multiple access scheme, namely time-reversal multiple access (TRMA). Results demonstrate the superiority of our scheme, in terms of the bit error rate (BER), particularly in sparse multi-path environments compared to TRMA, with a percent of improvement in the average BER between 73.5%−98.31%. This improvement is also accompanied by a reduced overhead compared to traditional TRMA, which necessitates users' channel state information feedback to the base station for TR precoding. Moreover, our findings indicate that at high signalto-noise ratio values, the diversity gain for a particular user is proportional to the product of the user's order, determined by its channel strength, and the number of its channel taps.