FIGURE 1. (a) The proposed 5G Traffic-GAN model architecture. (b) Temporal block structure of discriminator.
FIGURE 2. Discriminator network architecture
FIGURE 3. Generator network architecture
The following should be installed.
Window OS V11
WSL V2
Docker desktop for window
docker image : eungbean/deepo:lab
Nvidia Driver : 510.06
CUDA : 11.6
CUDNN : 7.6.5
I trained models in this computer environment.
CPU : Ryzen 7 1700
RAM : 16GB
GPU : Geforce GTX 1070 (6G)
Train time
1 epoch : 565 sec
665 epoch (best performance) : 104 hour
You can install modules related to the current project with the following items.
pip install tensorboardX
> cd your_path/TGSC_LSTM-TCN-GANs
> python3 main.py