Radio Signal Classification

Machine Learning Dataset for Radio Signal Classification

This is a dataset of radio signals of different waveforms as they occur primarily in the HF bands. The data has been created synthetically by applying a AWGN + Watterson Fading (to account for ionospheric propagation) channel model, plus random frequency and phase offset. The dataset enables experiments on signal and modulation classification using modern machine learning such as deep learning with neural networks.

Related Publications:

  • Stefan Scholl: Classification of Radio Signals and HF Transmission Modes with Deep Learning, 2019,
  • Stefan Scholl (DC9ST): Classification of shortwave radio signals with deep learning, Software Defined Radio Academy 2021

The dataset has the following properties:

  • 172,800 signal vectors
  • each signal vector has 2048 complex IQ samples with fs = 6 kHz
  • signal power is normalized to 1
  • frequency offset: +- 250 Hz
  • random phase offset
  • SNR values: 25, 20, 15, 10, 5, 0, -5, -10 dB
  • fading channel: Watterson Model as defined by CCIR 520
  • 18 Transmission Modes / Modulations: Morse, PSK31, PSK63, QPSK31, RTTY 45/170, RTTY 50/170, RTTY 100/850, Olivia 8/250, Olivia 16/500, Olivia 16/1000, Olivia 32/1000, DominoEx 11, MT63/1000, Navtex, USB audio, LSB audio, AM audio, HF fax

The dataset is available for download in 2-D numpy array format with shape=(172800, 2048)

Recent Posts