Multiple-Antenna Cooperative Spectrum Sensing Based on the Wavelet Transform and Gaussian Mixture Model.

Area of Science:

• Wireless Communication
• Signal Processing
• Cognitive Radio

Background:

• Spectrum sensing is crucial for efficient cognitive radio (CR) operation.
• Traditional methods face challenges with channel loss and hidden terminals, impacting secondary user (SU) performance.
• Existing techniques often require complex threshold derivations.

Purpose of the Study:

• To propose a multiple-antenna cooperative spectrum sensor (MAWG) for improved CR systems.
• To enhance SU spectrum sensing performance, particularly under adverse channel conditions.
• To introduce a novel sensing data fusion method that avoids complex threshold calculations.

Main Methods:

• Utilizing wavelet transform for noise reduction in signals collected by multiple-antenna SUs.
• Employing a fusion center (FC) to extract statistical features from pre-processed signals.
• Implementing a Gaussian mixture model (GMM) for classifier training based on a two-dimensional feature vector derived from clustered SUs.

Main Results:

• The MAWG method effectively reduces signal noise through wavelet transform.
• A novel data fusion approach extracts a two-dimensional feature vector, simplifying decision-making.
• The GMM classifier, trained on this feature vector, successfully performs spectrum sensing without explicit threshold derivation.
• Simulations in the κ-μ channel model demonstrate MAWG's effectiveness in improving spectrum sensing.

Conclusions:

• The proposed MAWG method offers a robust solution for cooperative spectrum sensing in CR systems.
• Wavelet transform and GMM-based fusion significantly enhance sensing accuracy and efficiency.
• MAWG overcomes limitations of traditional methods, improving SU performance in challenging environments.