TensorFlow implementation of the paper: Universal Function Approximation Through Over-the-air Computing: A Deep Learning Approach
The scripts MSE_vs_SNR_flat_fading.py and MSE_vs_users_flat_fading.py can be used to reproduce the results of the paper. Some straightforward modifications of the scripts may be needed to generate all figures. Additionally, while the script MSE_vs_SNR_selective_fading.py was not utilized in the paper, it is functional and can be employed for the case of selective frequency fading channels. Finally, note that the device's transmit power has been set equal to 1W in the simulation settings of the paper, and thus, does not explicitly appear in the scripts.
If you find the code useful, please cite our paper.
@ARTICLE{10506799,
author={Bouzinis, Pavlos S. and Evgenidis, Nikos G. and Mitsiou, Nikos A. and Tegos, Sotiris A. and Diamantoulakis, Panagiotis D. and Karagiannidis, George K.},
journal={IEEE Open Journal of the Communications Society},
title={Universal Function Approximation Through Over-the-Air Computing: A Deep Learning Approach},
year={2024},
volume={},
number={},
pages={1-1},
doi={10.1109/OJCOMS.2024.3392508}}