Automatic Classification of 5G Waveform-Modulated Signals Using Deep Residual Networks

Area of Science:

• Wireless Communication Systems
• Signal Processing
• Machine Learning

Background:

• Modulation identification is critical for 5G and future wireless networks employing diverse multicarrier waveforms.
• Existing methods struggle with the complexity and variety of advanced 5G signal types.

Purpose of the Study:

• To develop an innovative Automatic Modulation Classification (AMC) algorithm for advanced 5G waveforms.
• To leverage deep learning, specifically Deep Residual Networks (DRNs), for enhanced modulation recognition.

Main Methods:

• An Automatic Modulation Classification (AMC) algorithm utilizing a Deep Residual Network (DRN) architecture.
• Integration of Principal Component Analysis (PCA) for dimensionality reduction and feature refinement.
• Classification of complex 5G waveforms including OFDM, FOFDM, FBMC, UFMC, and WOLA with 16-QAM and 64-QAM.

Main Results:

• The proposed DRN-based model demonstrates significantly improved classification accuracy and robustness compared to traditional machine learning approaches.
• Achieved high performance in classification recall, precision, accuracy, and F-measure for diverse 5G waveforms.
• This represents the first application of deep learning for classifying such a comprehensive set of 5G waveforms.

Conclusions:

• Deep Residual Networks (DRNs) are highly effective for Automatic Modulation Classification (AMC) in complex wireless environments.
• The developed algorithm enhances adaptive signal processing capabilities for future wireless communication technologies.
• The study validates the potential of deep learning in advancing modulation recognition for next-generation networks.