Automatic Modulation Classification of Digital Communication Signals Using SVM Based on Hybrid Features,...

Area of Science:

• Digital Signal Processing
• Machine Learning for Communications

Background:

• Digital communication signals are crucial in radio and underwater systems.
• Modulation classification is vital but challenging due to poor channel conditions and low signal-to-noise ratio (SNR).

Purpose of the Study:

• To propose a novel method for automatic modulation classification of digital communication signals.
• To enhance classification performance and noise tolerance.

Main Methods:

• A support vector machine (SVM) classifier was employed.
• Hybrid features combining cyclostationary properties and information entropy were utilized.
• Three new features were introduced: maximum normalized cyclic spectrum value (non-zero cyclic frequency), Shannon entropy of the cyclic spectrum, and Renyi entropy of the cyclic spectrum.

Main Results:

• The proposed method demonstrated superior classification performance compared to existing techniques.
• The new features exhibited strong anti-noise capabilities without requiring prior signal information.
• An 'one against one' SVM strategy was implemented for classification.

Conclusions:

• The novel hybrid feature-based SVM method offers improved automatic modulation classification.
• The approach is robust against noise and suitable for challenging communication environments.
• This method advances digital signal processing for reliable communication systems.