A Survey of Blind Modulation Classification Techniques for OFDM Signals.

Area of Science:

• Electrical Engineering
• Computer Science
• Signal Processing

Background:

• Blind modulation classification (MC) is essential for adaptive transceivers in future wireless communications.
• MC supports 6G systems by enhancing spectral and power efficiency and reducing latency.
• It is a key component for intelligent software-defined radios (SDR).

Purpose of the Study:

• To systematically review various MC techniques for orthogonal frequency division multiplexing (OFDM) signals.
• To compare the advantages and limitations of statistical and machine learning (ML) based MC methods.
• To provide insights into practical experimental works and future research directions.

Main Methods:

• Focus on statistical methods: likelihood-based (LB), maximum a posteriori (MAP), and feature-based (FB).
• Focus on ML methods: k-nearest neighbors (KNN), support vector machine (SVM), decision trees (DTs), convolutional neural networks (CNNs), recurrent neural networks (RNNs), and long short-term memory (LSTM).
• Simulation of primary statistical and ML algorithms under various constraints for fair comparison.

Main Results:

• Comparison of characteristics, advantages, and disadvantages of different MC techniques.
• Performance evaluation in terms of bit error rate (BER) for MC.
• Survey of practical experimental results using National Instrument hardware in indoor environments.

Conclusions:

• The survey provides a comprehensive overview of blind MC techniques for OFDM signals.
• It highlights the trade-offs between different statistical and ML approaches.
• Identifies open problems and future research directions in blind MC.