Joint Demodulation and Decoding with Multi-Label Classification Using Deep Neural Networks
Imtiaz Ahmed, Wenjie Xu, Ramesh Annavajjala, Wook-Sung Yoo · 2021
In this paper, we leverage the power of artificial intelligence in the receiver design for joint baseband demodulation and channel decoding. We consider a point-to-point communication system and develop a deep neural network (DNN) based joint demodulator and decoder (DeModCoder) that accomplishes the tasks of demodulation and decoding in a single operational block. We incorporate a multi-label classification (MLC) scheme for the considered DNN framework, which is trained offline over a wide-range of signal-to-noise ratios (SNRs) in a supervised learning manner and deployed online in real-time applications. Simulation results demonstrate that our developed DeModCoder outperforms the conventional block-based sequential demodulation and decoding schemes. We also observe that the MLC DeModCoder shows better performance than conventional multiple output classifier in high SNR region while incurring lower computational complexity.