Learning environment-aware and hardware-compatible mmWave beam-forming codebooks using Artificial Neural Networks (ANN). Please refer to Learning Beam Codebooks with Neural Networks: Towards Environment-Aware mmWave MIMO for more details.
1-Keras with Theano backend.
2-Numpy library.
3-Python 3.6 or above.
Optional:
1- MATLAB for data preperation.
The implementation of the complex-valued fully-connected layer is guided by that in Deep Complex Nets with some modifications.
If you use this code or part of it, please cite the following paper:
@article{zhang2020learning,
title={Learning Beam Codebooks with Neural Networks: Towards Environment-Aware mmWave MIMO},
author={Zhang, Yu and Alrabeiah, Muhammad and Alkhateeb, Ahmed},
journal={arXiv preprint arXiv:2002.10663},
year={2020}
}