Multilayered SDN security with MAC authentication and GAN-based intrusion detection.

Area of Science:

• Cybersecurity
• Computer Networks
• Deep Learning

Background:

• Software-defined networks (SDN) are increasingly targeted by cyberattacks due to their role in 5G data transmission.
• Existing intrusion detection systems often suffer from low accuracy and high false positive rates, necessitating advanced solutions.

Purpose of the Study:

• To develop a robust, multilayered intrusion detection system for SDN environments.
• To enhance network security and efficiency through advanced authentication and deep learning techniques.

Main Methods:

• Implementation of a novel four-Q curve authentication system utilizing elliptic curve cryptography for secure and efficient authentication.
• Application of univariate ensemble feature selection for optimal switch selection.
• Utilizing a Dual Discriminator Conditional Generative Adversarial Network (DDcGAN), optimized by the Sheep Flock Optimization Algorithm (SFOA), for classifying network traffic.
• Employing the Growing Self-Organizing Map (GSOM) for categorizing suspicious packets.

Main Results:

• The DDcGAN-based intrusion detection system achieved a high accuracy of 98.29% and an F1 score of 0.975.
• Demonstrated superior performance over state-of-the-art methods in precision, sensitivity, and reduced false-positive rates (2.05%).
• Achieved a true positive rate of 99.04% even with 50% malicious nodes and reported 4.5% energy savings.

Conclusions:

• The proposed four-Q curve authentication and DDcGAN-based intrusion detection system offers a significant advancement in securing SDN environments.
• The system effectively balances high detection accuracy with computational efficiency and reduced energy consumption.
• This research provides a promising framework for future cybersecurity solutions in advanced network infrastructures.