GRU-SVM Based Threat Detection in Cognitive Radio Network

Area of Science:

• Computer Science
• Electrical Engineering
• Telecommunications

Background:

• Cognitive radio networks (CRN) face significant security challenges during spectrum sensing, with malicious users degrading network performance.
• Machine learning and deep learning offer promising solutions for detecting network anomalies and threats in CRN.
• Gated recurrent units (GRU), a type of deep learning model, are underutilized in CRN despite their effectiveness with smaller datasets.

Purpose of the Study:

• To investigate the application of gated recurrent units (GRU) for spectrum sensing in cognitive radio networks.
• To combine GRU with a support vector machine (SVM) classifier to enhance the detection of malicious users.
• To evaluate the performance of the proposed GRU-SVM model in identifying authorized and unauthorized users.

Main Methods:

• Utilized gated recurrent units (GRU), a lightweight variant of LSTM, for training and testing spectrum sensing data.
• Employed a support vector machine (SVM) classifier in the output layer for distinguishing between authorized and malicious users.
• Developed a novel combined model integrating GRU and SVM specifically for cognitive radio network applications.

Main Results:

• Achieved a high testing accuracy of 82.45% and a training accuracy of 80.99%.
• Demonstrated a perfect detection probability of 1 for malicious users.
• The proposed GRU-SVM model showed effectiveness at 65 epochs.

Conclusions:

• The integration of GRU and SVM presents a novel and effective solution for enhancing security in cognitive radio networks.
• GRU models offer a computationally efficient approach for deep learning applications in CRN, particularly with limited data.
• The proposed method significantly improves the detection of malicious users, thereby bolstering the overall performance and reliability of CRN.