Packet Format Detection and Modulation Classification of Wireless LAN Using Distributed Convolutional Neural Network

Abstract

This paper proposes a method to detect the packet format and classify signal modulation type of wireless LAN signals using the distributed model of Convolutional Neural Network (CNN). The main advantages of this method compared to the conventional are the high accuracy of the model and the flexibility for lowering the complexity. These are achieved because the CNN models are trained to perform multiple small classification tasks instead of a single big classification task. This method makes retraining much easier. The five-fold cross-validation is applied to assess the performance for training and testing the model. To validate the model, actual Wi-Fi signals are used to test the model using Software Define Radio (SDR). The result shows the proposed method can accurately classify different packet formats and signal modulation types above 90%, even when different timing offsets occur.