Statistical Feature-Aided TCN-GRU Network for Radio Signal Modulation Classification

KAYA Z., IŞIK Ş.

9th International Artificial Intelligence and Data Processing Symposium, IDAP 2025, Malatya, Turkey, 6 - 07 September 2025, (Full Text) identifier

  • Publication Type: Conference Paper / Full Text
  • Doi Number: 10.1109/idap68205.2025.11222285
  • City: Malatya
  • Country: Turkey
  • Keywords: AMC, RadioML 2016.10a, SNR, TCN-GRU
  • Bilecik Şeyh Edebali University Affiliated: Yes

Abstract

In this study, we present the Automatic Modulation Classification (AMC) under varying Signal-to-Noise Ratio (SNR) conditions. A total of 11 different radio signals with 21 different values of the signal-to-noise ratio (SNR), ranging from -20 decibels (dB) to +18 dB, are analyzed. The RadioML 2016.10a dataset, containing 220,000 samples, is used in the experiments, with 80% allocated for training and 20% for testing. A set of 53 statistical features is derived from the In-phase (I) and Quadrature (Q) signal channels. These features are used to train a hybrid deep learning model based on Temporal Convolutional Networks (TCNs) and Gated Recurrent Units (GRUs). The proposed model achieves an overall classification accuracy of 56% across all SNR levels, with the highest accuracy of 89% obtained at an SNR of +12 dB.