Classification of Channel Encoders Using Convolutional Neural Network

Abstract

In digital communication systems, channel encoders play a crucial role in rectifying random errors introduced by the channel. Typically, information about the type and parameters of channel encoders used at the transmitting end is available at the receiver. However, in non-cooperative scenarios like military communication systems, encoder types and parameters may be only partially known or entirely unknown. This paper explores the feasibility of employing a deep learning approach to classify four different types of encoders: block, convolutional, Bose-Chaudhuri-Hocquenghem (BCH), and polar encoders. Utilizing a convolutional neural network (CNN) model for classification, our proposed approach achieves classification accuracy exceeding 95% upto bit-error-rate (BER) value of 0.03. The results also indicate that the accuracy improves with the input sample length. © 2024 IEEE.