Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Digital Operating Mode Classification of Real-World Amateur Radio Transmissions

Maximilian Bundscherer, Thomas H. Schmitt, Ilja Baumann, Tobias Bocklet

TL;DR

This work tackles automatic modulation classification (AMC) of amateur-radio digital operating modes under real-world channel conditions. It trains spectrogram-based vision models (CNNs and a Vision Transformer) on limited non-transmitted data augmented online to simulate impairments, and evaluates on real SDR-recorded transmissions. EfficientNetB0 achieves $93.80\%$ accuracy across 17 operating modes and $85.47\%$ across 98 parameterized signals, with longer signal durations yielding stronger performance and robustness to SNR variations. The study offers practical, plug-and-play insights for spectrum monitoring using standard SDR/WebSDR setups in regulatory and operational contexts.

Abstract

This study presents an ML approach for classifying digital radio operating modes evaluated on real-world transmissions. We generated 98 different parameterized radio signals from 17 digital operating modes, transmitted each of them on the 70 cm (UHF) amateur radio band, and recorded our transmissions with two different architectures of SDR receivers. Three lightweight ML models were trained exclusively on spectrograms of limited non-transmitted signals with random characters as payloads. This training involved an online data augmentation pipeline to simulate various radio channel impairments. Our best model, EfficientNetB0, achieved an accuracy of 93.80% across the 17 operating modes and 85.47% across all 98 parameterized radio signals, evaluated on our real-world transmissions with Wikipedia articles as payloads. Furthermore, we analyzed the impact of varying signal durations & the number of FFT bins on classification, assessed the effectiveness of our simulated channel impairments, and tested our models across multiple simulated SNRs.

Digital Operating Mode Classification of Real-World Amateur Radio Transmissions

TL;DR

This work tackles automatic modulation classification (AMC) of amateur-radio digital operating modes under real-world channel conditions. It trains spectrogram-based vision models (CNNs and a Vision Transformer) on limited non-transmitted data augmented online to simulate impairments, and evaluates on real SDR-recorded transmissions. EfficientNetB0 achieves accuracy across 17 operating modes and across 98 parameterized signals, with longer signal durations yielding stronger performance and robustness to SNR variations. The study offers practical, plug-and-play insights for spectrum monitoring using standard SDR/WebSDR setups in regulatory and operational contexts.

Abstract

This study presents an ML approach for classifying digital radio operating modes evaluated on real-world transmissions. We generated 98 different parameterized radio signals from 17 digital operating modes, transmitted each of them on the 70 cm (UHF) amateur radio band, and recorded our transmissions with two different architectures of SDR receivers. Three lightweight ML models were trained exclusively on spectrograms of limited non-transmitted signals with random characters as payloads. This training involved an online data augmentation pipeline to simulate various radio channel impairments. Our best model, EfficientNetB0, achieved an accuracy of 93.80% across the 17 operating modes and 85.47% across all 98 parameterized radio signals, evaluated on our real-world transmissions with Wikipedia articles as payloads. Furthermore, we analyzed the impact of varying signal durations & the number of FFT bins on classification, assessed the effectiveness of our simulated channel impairments, and tested our models across multiple simulated SNRs.
Paper Structure (11 sections, 2 equations, 2 figures, 3 tables)

This paper contains 11 sections, 2 equations, 2 figures, 3 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Data
  4. Radio Signal Generation & Overview
  5. Radio Signal Transmission, Recording & Postprocessing
  6. Methods
  7. Classification Models
  8. Online Data Augmentation & Channel Impairments
  9. Experiments & Results
  10. Discussion
  11. Conclusion

Figures (2)

  • Figure 1: This normalized confusion matrix visualizes the operating mode (OM) confusions of our best model, EfficientNet-B0 (EN-B0), considering a duration of $2$s and $128$ FFT bins. Evaluated for both receivers R0 and R1.
  • Figure 2: The operating mode parameters (OMP) classification accuracies of our best model, EfficientNet-B0 (EN-B0), utilizing $128$ FFT bins across multiple simulated SNRs.