Differential Deep Detection in Massive MIMO With One-Bit ADC

TL;DR

This paper introduces a differential detection scheme for massive MIMO systems with one-bit ADCs, utilizing the Bussgang theorem and deep learning to improve amplitude and phase detection, leading to enhanced spectral efficiency.

Contribution

It develops a novel differential detection method for one-bit massive MIMO systems, including a deep neural network for amplitude detection, improving spectral efficiency over coherent systems.

Findings

The phase detector successfully decodes differential phase information.

The neural network improves amplitude detection accuracy.

Spectral efficiency surpasses that of one-bit coherent systems.

Abstract

This article presents a differential detection scheme for the uplink of a massive MIMO system that employs one-bit quantizers on each receive antenna. We focus on the detection of differential amplitude and phase shift keying symbols and we use the Bussgang theorem to express the quantized received signal in terms of quantized signals received during previous channel uses. Subsequently, we derive the maximum likelihood detector for the differentially encoded amplitude and phase information symbols. We note that while the one-bit detector can decode the differentially encoded phase information symbols, it fails to decode the differentially encoded amplitude information. To decode the amplitude information, we present a one-bit variable quantization level (VQL) system and train a deep neural network to perform two-symbol differential amplitude detection. Through Monte-Carlo simulations, we empirically validate the performance of the proposed amplitude and phase detectors. The presented numerical results show that the spectral efficiency attained in one-bit differential systems is better than the spectral efficiency attained in one-bit coherent systems