An official repository for the code for the paper:
Nikita Smirnov and Sven Tomforde: Real-time rate control of WebRTC video streams in 5G networks: Improving quality of experience with Deep Reinforcement Learning. Journal of Systems Architecture, vol. 148, 2024. Link: https://www.sciencedirect.com/science/article/pii/S1383762124000031
Pull docker image from Docker Hub:
docker pull gehirndienst/gymir5g_arcs23:latest
then run a container:
docker run -it --rm --name gymir5g_arcs23 gehirndienst/gymir5g_arcs23:latest bash
attach vs-code to the container and install vs-code python and jupyter extensions. Selected pre-installed poetry environment as python interpreter
Gymir5G is still under the active development and its source code is not yet published. However, there is a pre-built execution file of the latest snapshot to the date of paper submission (16.10.2023) that can be used in a container to run the simulation.
The full code for python part is presented in gymirdrl folder.
This folder contains the code, which was used to test the trained Soft Actor-Critic model for the paper (/arcs23). It is copied to the container and is located in /opt/omnetpp-workspace/gymir5g/experiments/arcs23 folder.
logs_plots folder contains the experiments results and plots referenced in ARCS23 paper. It also contains the Jupyter notebook (plot.ipynb) used to generate the plots. training_logs folder contains the logs, metric, tb and omnetpp logs from the training. env_cfg.json is the configuration file for the environment. hparam_cfg.json is the configuration file for the hyperparameters used in the training. model.zip is the trained sb3 SAC model. paper.ipynb is the jupyter notebook with the code used to test the models and GCC on 3 validation scenarios mentioned in ARCS23 paper. NOTE: works only in the above mentioned container. stream_cfg.json is the configuration file for the stream used in the experiments. NOTE: it contains some not-omitted parameters that are not used in the experiments for ARCS23 paper.
This folder contains the C++ code for the GCC (Google Congestion Control) implementation used in the experiments for ARCS23 paper. Based on https://github.com/thegreatwb/HRCC/.
GPLv3 (see LICENSE file)
Nikita Smirnov, Sven Tomforde Department of Computer Science, University of Kiel, Germany
If you have any questions, please contact Mr. Nikita Smirnov ([email protected])