Gaussian Two-Way Channels With Discrete Inputs and Quantized Outputs

TL;DR

This paper investigates Gaussian two-way channels with quantized outputs, determining optimal quantizers, analyzing constellation-based transmitters, and demonstrating how constellation rotation can enlarge the achievable communication region.

Contribution

It introduces a numerical approach to optimize uniform quantizers and develops a formulation for the Shannon achievable region in two-way channels with discrete inputs.

Findings

Optimal uniform quantizers depend on SNR levels.

Constellation rotation enlarges the achievable region.

Numerical results demonstrate the benefits of constellation-based strategies.

Abstract

In this paper, Gaussian two-way channel with uniform output quantization is studied. For Gaussian inputs, the optimum uniform finite-level quantizer is determined numerically for different values of Signal-to-Noise Ratio (SNR). The two-way channel with constellation-based transmitters is then investigated. A formulation for the so-called Shannon achievable region of this channel is developed and numerical computations of this region are presented for particular constellations. It is shown that if one transmitter utilizes a rotated version of the constellation used at the other transmitter, the Shannon achievable region can be enlarged