This code was used in the following unpublished work (preprint not available yet).
[1] Steffen Gracla, Alea Schröder, Carsten Bockelmann, Armin Dekorsy, "Model-free Robust Beamforming in Satellite Downlink using Reinforcement Learning", under review.
Email: {gracla, schroeder, bockelmann, dekorsy}@ant.uni-bremen.de
The code version associated with this paper along with the used learned models and evaluation results is found in the releases. The project structure is as follows
.
├── models | trained models
├── outputs
│ ├── metrics | metrics from evaluation
├── README.md | this file
├── requirements.txt | project dependencies
├── src | python source related to..
│ ├── analysis | evaluation
│ ├── config | configuration
│ ├── data | data generation, i.e., satellite model
│ ├── models | learning models
│ ├── plotting | plotting
│ ├── tests | code tests
└── └── utils | shared helper functions