Implementation of ITU-T Recommendation G.1072

This repo provides the implementation of ITU-T G.1072 "Opinion model predicting gaming quality of experience for cloud gaming services" based on network and comperession parameters.

• The G.1072 Planning model that can predict the cloud gaming quality based on the video paramters, bitrate, framerate, encoding resolution, gaming video complexity and network parameters, packetloss, delay on MOS scale from 1 to 5 (5 best, 1 worse).

How to Use

In order to run the code please use run it based on the parameters of the model as follows:

• Test the model based on the parameters, which requires specifying all video parameters, bitrate, framerate, ecoding resolution (width x height,e.g. 1920x1080) and video complexity (Vcomplexity) level ('High','Medium' ,'Low') and network parameters packetloss (if freezing is the effect), packetlossUDP (if slicing is the packet loss effect), delay and delay (interaction) complexity (Icomplexity) . To do so, you can run it as the following example:

python3 G1072.py --bitrate=50  --framerate=60 \
                --packetloss=0.20 --packetlossUDP=0 \
                --delay=0 --coding_res=1920x1080 \
                --Icomplexity=High --Vcomplexity=High

For more help run:

python3 G1072.py -h

Application Range of Parameters

Please note that the model only works based on the range of parameters used for training the model.

• Bitrate: 0.3 kbps to 50 Mbps
• Framerate: 10 fps to 60 fps
• Resolution: 1920x1080, 1280x720 or 640×480
• Video Complexity Class: High, Medium, Low
• Interaction Sensitivity Class: High, Medium, Low
• Packetloss (Freezing effect): 0 - 5 (it will be seen as 0 to 5%)
• Packetloss (Slicing effect): 0 - 2 (it will be seen as 0 to 2%)
• Delay: 0 - 400

Output of the model

The model gives you four estimations:

• Overall Quality
• Interaction Quality (known as Input Quality in G,1072) due to delay
• Interaction Quality (known as Input Quality in G,1072) due to packet loss (freezing effect/frame loss)
• Video Quality based on refitted ITU-T Recommendation G.1071
• Video Fragementation (based on https://github.com/stootaghaj/GamingVQA)
• Video Unclearness (based on https://github.com/stootaghaj/GamingVQA)

Example

Here you can find an example for a condition with 2 Mbps, 60 fps, packet loss of 0%, delay of 200 ms, Interaction complexity of "High", and Video Complexity of "High" level:

python3 G1072.py --bitrate=2  --framerate=60 \
                --packetloss=0 --packetlossUDP=0 \
                --delay=200 --coding_res=1920x1080 \
                --Icomplexity=High --Vcomplexity=High

Output:

Overal Quality: 1.703640262361372
Interaction Quality (delay): 3.1691042828686484
Interaction Quality (packetloss): 4.64
Video Quality based on refitted G.1071: 2.8929607456164215
Video Unclearness: 3.6660390586801817
Video Fragmentation: 2.852382320395556

Prepration

Install python3 and pip, if they are not already installed. Follow the platform specific installation instructions. The following step should be performed to prepare the setup.

    git clone https://github.com/stootaghaj/ITU-G1072.git
    cd ITU-G1072
    pip3 install -r requirements.txt

Testing

Run Pytest:

pip3 install -r requirements.dev.txt
python3 -m pytest test/test.py

Citation

Please cite the ITU-T Recommendation G.1072 if you use the code or to get more insight about the model:

    @inproceedings{g1072,
    title={ITU-T Recommendation G.1072: Opinion model predicting gaming quality of experience for cloud gaming services},
    booktitle={International Telecommunication Union},
    year={2020},
    organization={ITU}
    }

Link to ITU-T Recomendation G.1072: https://www.itu.int/rec/T-REC-G.1072-202001-I/en

Contributors

The work has been done within the ITU-T Study Group 12 under work item, G.OMG. Many researchers contributed to this work from TU Berlin, Simula Research Lab, Kingston University, Deutsche Telekom. The current code is made available to test the model. It has to be noted that code is different than the original code and might include bugs. Please contact saman.zadtootaghaj@qu.tu-berlin.de if any issues is found.

• Steven Schmidt
• Saman Zadtootaghaj
• Saeed Shafiee Sabet
• Nabajeet Barman

License

For a commercial license, you must contact the respective rights holders of the standards ITU-T Rec. G.1072, ITU-T Rec. G.1071.

Permission is granted, free of charge, to use the software for non-commercial research purposes.