UWB Receiver via Deep Learning in MUI and ISI Scenarios

In this paper, we consider a multiuser impulse radio ultra-wideband (UWB) system and focus on a single target user's signal reception in the presence of multiuser interference (MUI) and/or inter-symbol interference (ISI). We propose a deep learning (DL) based receiver, which is trained off-line by simulated data to learn the joint effect of UWB pulsing, channel effect, and signal detection. Then we apply it online to recover the real-time transmitted data symbols. Numerical results demonstrate that the deep-learning based receiver can efficiently learn such joint effect in the presence of MUI and ISI, leading to much better bit error rate (BER) performance than conventional correlation receiver and other existing receivers.