Automatic modulation classification works well over free-space sub 6 GHz links
Sub-6 GHz Over-the-Air AMC via Curriculum Fine-Tuned CNN-Transformers
Machine Learning
Summary
It can be hard to tell which type of radio signal modulation is being used when the signals travel through the air, especially because real-world effects like path loss and antenna misalignment can change the signal. The authors studied a machine learning model that combines convolutional neural networks and transformers, training it first on controlled data and then fine-tuning it on real over-the-air signals at different distances and antenna alignments. They found that the model can still identify signal types with high accuracy even under these realistic conditions. Their work helps show how such models perform beyond ideal lab setups.
What this means in practice
- •For wireless communication engineers: Deploy AMC models that remain accurate over real free-space links despite antenna misalignment and path loss at sub-6 GHz frequencies.
- •For signal intelligence analysts: Use fine-tuned CNN-Transformer models for reliable radio signal classification in operational over-the-air scenarios.
Authors
Nurettin Safak, Muhammet Sefa Demirel, Alperen Marasli, Taha Eren Atmaca, Durdu Can Yerdeyatar, Ozgun Ersoy
Abstract
Automatic modulation classification (AMC) models are frequently trained and validated on synthetic or channel-cabled data, leaving open the question of how they behave once path loss and antenna pointing error are introduced by a genuine free-space link. We report a curriculum fine-tuning study of a hybrid CNN-Transformer AMC model. The general-purpose, all-32-class dataset underlying the model was built entirely at 915 MHz, on a controlled, clock/PPS-synchronized MIMO-expansion-cable link (not spatial-multiplexing transmission); all subsequent free-space, real-hardware experimentation - the sequential fine-tuning curriculum, matched-distance evaluation, and every reported over-the-air accuracy figure - was carried out at 4 GHz, across five directional-antenna distances (25, 35, 50, 70, 75 cm) under fixed TX/RX gain. Three distances used near-ideal antenna alignment (~99%) and two used a deliberately introduced partial misalignment (~85%), fine-tuned last in the curriculum. We report matched-distance test accuracy (91.8-93.7% across all five 4 GHz conditions) and confusion-matrix analysis grounded in RF theory, and report honestly where our sequential fine-tuning order confounds cumulative link adaptation with antenna alignment, rather than overstating what the data can support.