Related Jira project: ICAT-3256
MWI, 2024
CODEOWNERS: pueyo99@gmail.com, daniel.camps@i2cat.net, miguel.catalan@i2cat.net,
MANTAINERS: miguel.catalan@i2cat.net
Python implementation of PHAUL, a Deep Reinforcement Learning Agent that produces the optimal flow allocations for Integrated Access Backhaul networks.
- Ubuntu >= 20.04
- Python >= 3.7
- TensorFlow >= 2.6
- OpenAI Gym
- Stable-Baselines3
- conda
After cloning the repository to local, install it using conda:
conda env create -f environment.yml
To change the parameters of the agent modify the src/cfg.yaml file, where you can find all the available parameters.
To run the agent execute python src/main.py. The src/main.py file is configured to run multiple executions of the agent in parallel, using as many CPUs as required, and varying the values of the parameters used. The results of the executions are stored as .txt files. The script src/utils/results_to_csv.py can be used to generate a single CSV file contatining all the results of a single execution.
This code has been developed within the research project NANCY. This project has received funding from the European Union’s Horizon 2022 research and innovation programme under grant agreement No 101096456.
This project started on 1 January 2023 and will end on 31 December 2025, funded under HORIZON-JU-SNS-2022 Framework Programme, with an overall budget of €6447428,75 and an EU contribution of €5999798,00, coordinated by PANEPISTIMIO DYTIKIS MAKEDONIAS. More information about the grant at https://cordis.europa.eu/project/id/101096456
This code has been developed by Fundació Privada Internet i Innovació Digital a Catalunya (i2CAT). i2CAT is a non-profit research and innovation centre that promotes mission-driven knowledge to solve business challenges, co-create solutions with a transformative impact, empower citizens through open and participative digital social innovation with territorial capillarity, and promote pioneering and strategic initiatives. i2CAT aims to transfer research project results to private companies in order to create social and economic impact via the out-licensing of intellectual property and the creation of spin-offs. Find more information of i2CAT projects and IP rights at https://i2cat.net/tech-transfer/
This code is licensed under the terms AGPLv3. Information about the license can be located at link. If you find that this license doesn't fit with your requirements regarding the use, distribution or redistribution of our code for your specific work, please, don’t hesitate to contact the intellectual property managers in i2CAT at the following address: techtransfer@i2cat.net