Blind Interleaver Classification Using Deep Learning Techniques over Rayleigh Fading

Channel encoders and interleavers are crucial for correcting random and burst errors introduced by noisy channels in digital communication systems. Typically, information about the types of channel encoders and interleavers, along with their parameters used at the transmitting end, is available at the receiver. However, in military communication systems, the encoder/interleaver type and parameters are often partially known or unknown. In reconfigurable receivers and adaptive modulation and coding (AMC)-based systems, blind identification of channel encoders and interleavers enhances spectral efficiency. This paper explores the use of a deep learning approach to identify three different types of interleavers namely block, convolutional, and helical, under Rayleigh fading conditions. We propose a hybrid convolutional neural network (CNN)- support vector machine (SVM) model for classification, with input data consisting of block-encoded and convolutional-encoded data. Our hybrid CNN-SVM model achieves over $\mathbf{9 5 \%}$ classification accuracy across varying signal-to-noise ratio (SNR) values. The findings also show that accuracy improves with longer input sample lengths, albeit at the cost of increased training and testing time. Finally, our proposed model demonstrates superior classification accuracy compared to the CNN model.