The Matlab source codes are provided to supplement our research paper "Feedback Sample Selection Methods Allowing Lightweight Digital Predistorter Adaptation". They allows to reproduce simulation results in our paper.
We have run simulations on Ubuntu OS, Matlab 2018a, but the simulations should be OS independent and all Matlab versions > 2018 should be compatible. Please note that for faster simulation executation Distributed Computation Toolbox is required, however, all simulations can be executed without parfor loops and hence should require no toolboxes.
The source codes are split into two parts: 1) calculation of results, 2) generating plots.
- run RUN_ANALYSIS_04.m to generate simulation results
- run PLOT_RESULTS_04.m to plot the simulation results
Please note that simulations are quite time demanding and can execute more than 24 hours (depending on the performance of HW).
By default, Matlab script RUN_ANALYSIS_04.m loads already optimised histograms from results_01_hist.mat. You can turn on histogram optimisation by changing the line 15 of RUN_ANALYSIS_04.m to "pars.is_hist_training = 1;".