Deep Learning-based Blind Identification of Channel Coding Schemes

The signal identification in blind radio frequency environment is gaining increasing importance in intelligent wireless communication systems. In this paper, we present a deep learning based blind identification of channel coding schemes. These include Convolutional coding, Turbo coding, and Low Density Parity Check coding-promising schemes for wireless communication systems. We have synthesized the dataset through MatLab simulations, where the randomly generated bits are encoded using specified coding schemes and modulated with BPSK and QPSK. The transmission of data has been simulated over a signal to noise ratio range of 0-20 dB in an AWGN channel, facilitating robust model training. Finally, a convolutional gated recurrent unit deep neural network is proposed. Unlike existing deep learning-based coding detection algorithms that handle only one modulation scheme, the proposed model processes information bits modulated with two different schemes. The method demonstrates average prediction accuracy approaching 87% to 90% over low regions of signal to noise ratio of the communication links.