Deep Learning Radio Frequency Signal Classification with Hybrid Images

Abstract

In recent years, Deep Learning (DL) has been successfully applied to detect and classify Radio Frequency (RF) Signals. A DL approach is especially useful since it identifies the presence of a signal without needing full protocol information, and can also detect and/or classify non-communication waveforms, such as radar signals. In this work, we focus on the different pre-processing steps that can be used on the input training data, and test the results on a fixed DL architecture. While previous works have mostly focused exclusively on either time-domain or frequency domain approaches, we propose a hybrid image that takes advantage of both time and frequency domain information, and tackles the classification as a Computer Vision problem. Our initial results point out limitations to classical pre-processing approaches while also showing that it's possible to build a classifier that can leverage the strengths of multiple signal representations.

A Radio Frequency Region-of-Interest Convolutional Neural Network for Wideband Spectrum Sensing

“…In the literature, we can find applications of deep neural networks, among others, in the classification of radio signals [ 5 , 6 ] or the use of DNNs as a detector in the basic processing band in the detection of signals with known characteristics [ 7 , 8 ]. However, there are very few approaches for the detection of signals in a wideband spectrum, returning proposals of regions of interest, i.e., center frequency and occupied bandwidth; and time stamps of occurrence of the signal of interest in the radio electromagnetic environment, i.e., de facto RPNs (region proposal network) for radio signals.…”

RFROI–CNN WSS – Radio Frequency Regions of Interests Convolutional Neural Network for Wideband Spectrum Sensing

“…In the literature, we can find applications of deep neural networks, among others, in the classification of radio signals [5], [6] or the use of DNNs as a detector in the basic processing band in the detection of signals with known characteristics [7], [8]. However, there are very few approaches for detection of signals in a wideband spectrum, returning proposal of regions of interest, i.e.…”