Automatic Classification of Wireless Fading Channels for UAV with CR Applications

Abstract

Unmanned aerial vehicle (UAV) with cognitive radio (CR) technology provides advantages such as coverage area, faster data rate, and improved traffic management. Conventional communication network suffers from ground reflections and to mitigate this, UAVs are used to serve the users. However, the non-line of sight paths in UAV based communication creates problems with channel effects. Hence it is essential to understand the type of fading that occurs in UAV based communication. Generally, the transmitted signal is affected by any one of the fading types such as flat fading, doubly selective fading, and frequency selective fading. In this work, wireless fading channels are classified using deep learning techniques. I and Q values of the modulation schemes are used as features for training and testing the models. Basic convolutional neural network (CNN) model and long short term memory (LSTM) are utilized for the classification and found that LSTM results better accuracy compared to basic CNN model. Computer simulations of proposed classification models are compared with conventional cumulants based classification results. � 2021 IEEE.