Multi - Task Learning with Convolutional Neural Network for Packet Format Detection and Modulation Classification of Wireless LAN

In this paper, we propose multi-task learning (MTL) using a convolutional neural network (CNN) to simultaneously perform packet format detection and modulation classification of wireless LAN (WLAN) signals. Compared to single-task learning, the main advantage of this method is the high classification accuracy and rapid learning with a lightweight neural network model. These are achieved because of learning shared knowledge (i.e., model weight or gradients) between related tasks. This shared representation model increases data efficiency and reduces overfitting. The WLAN signals generated by the Matlab wave generator are used to validate and test the model. The results show that even with different time offsets, the proposed method has an accuracy of 98% for packet format detection and 86% for modulation classification.