Performance Evaluation of Channel Decoding With Deep Neural Networks

Paper about the deep neural network decoder.

Introduction

We propose three types of NND, which build upon multi-layer perceptron (MLP), convolution neural network (CNN) and recurrent neural network (RNN).

Steps of Code

1. Generate the training samples saved at data/ under RNN/.

cd RNN/
python get_data.py

2. Run the deep neural network, BER is saved at result/.

python mlp.py
python cnn.py
python lstm.py

3. Plot the figure.

python plot.py