Wireless Standard Classification Using Convolutional Neural Networks

Open Set Wireless Signal Classification: Augmenting Deep Learning with Expert Feature Classifiers

“…Then, the FFT of each segment produces a high frequency resolution time-frequency signal representation. This feature was determined experimentally to be advantageous for CNN learning in [1].…”

“…The authors acknowledge Advanced Research Computing at Virginia Tech for providing computational resources that have contributed to the results reported within this paper, URL: https://arc.vt.edu/. Portions of this paper were presented at GLOBECOM 2021 [1], MILCOM 2021 [2], and MILCOM 2022 [3]. classification problems and have high computational complexity.…”

“…In the context of deep learning classifiers, data from "known" classes are In this paper, we propose an open set CNN-based wireless standard classifier that uses a modified version of the unknown class detector originally developed for text classification by the authors in [32] [3] by introducing expert feature classifiers and considering signal detection and isolation. Specifically, the deep learning feature preprocessing method from [1], the open set unknown class detector from [2], and the co-frequency signal classification system from [3] were used in this paper. The new contributions are as follows:…”

Deep Learning-Based Modulation Classification for OFDM Systems Without Symbol-Level Synchronization

“…The work in [16] studies modulation classification under multipath channel, but only deals with the single-carrier signals. The authors of [19] classify wireless signals, but recognizes wireless protocols, not modulations. There are papers on OFDM modulation classification without symbol-level synchronization based on mathematical modeling [20,21] without using DL.…”