Deep-Unfolding-For-Beamforming

This repository contains two scripts of code used to generate the results in research dealing with learn to rapidly and robustly optimize the beamforming problem in a hybrid architecture, as discussed in the paper attached to this repository. The first scenario uses a synthetic data denerated by python, and invovles i.i.d. entries to the channel matrix, denoted as H. On the contrary, scenario 2 deal with a Quadriga-generated data, a commonly used data-genratoe in the field of communications and signal processing. The generated data is attached to this repository, under the name "H_1000_2_8_7.mat", indicating its content: a thousand channel realizations, of 2 frequency bins, simulation a communication base station with 8 antennas, serving 7 users. The results can be obtained in more or less 6 hours. For convinence, the PDF version of each python-notebook is attached.