A Strong Machine Learning Classifier and Decision Stumps Based Hybrid AdaBoost Classification Algorithm for Cognitive...

Area of Science:

• Wireless Communications
• Machine Learning
• Signal Processing

Background:

• Cooperative spectrum sensing (CSS) is crucial for efficient spectrum utilization in cognitive radio networks (CRN).
• Traditional CSS methods face challenges in accurately detecting primary user behavior.
• Machine learning (ML) offers promising alternatives for enhancing CSS performance.

Purpose of the Study:

• To investigate ML-based CSS algorithms for CRNs.
• To propose a novel hybrid AdaBoost classification mechanism for improved primary user behavior pattern classification.
• To enhance the detection probability in spectrum sensing.

Main Methods:

• A hybrid AdaBoost classification mechanism combining a strong machine learning classifier (MLC) and decision stumps (DS).
• MLC is employed as the first-stage classifier, with DS as second-stage classifiers for spectrum energy vector classification.
• Simulations were conducted to evaluate the proposed algorithm's performance.

Main Results:

• The proposed hybrid AdaBoost algorithm demonstrated a higher detection probability.
• Performance was superior to conventional ML-based spectrum sensing algorithms.
• Outperformed conventional hard fusion-based CSS schemes.

Conclusions:

• The developed hybrid AdaBoost approach effectively classifies primary user behavior in CRNs.
• This method offers a significant improvement in detection probability for cooperative spectrum sensing.
• The proposed technique represents a viable advancement for cognitive radio network efficiency.