LSTM-based In-band full-duplex NOMA-OFDM Receiver with Non-linear SI Estimation/Cancellation

In this research, we aim to investigate Long Short Term Memory based SER performance in in-band full-duplex NOMA-OFDM receiver for detection of signal of interest from NOMA-OFDM multi users when both the weak user and strong user are present along with self-interference signal received at the receiving antenna from its own transmitting antenna of the base station operating in full-duplex mode under various channel imperfection. For this, we have proposed a three phase Long Short Term Memory (LSTM) based approach for canceling the non-linear self-interference in a full-duplex system and detecting the signal of interest. Extensive simulation studies show that the proposed LSTM-based method outperforms the existing deep learning as well as traditional MMSE-based methods.