Distributed Denial of Services (DDoS) attack detection in SDN using Optimizer-equipped CNN-MLP.

Area of Science:

• Computer Science
• Network Security
• Machine Learning

Background:

• Software-Defined Networks (SDN) offer enhanced network control but present unique security vulnerabilities.
• The centralized architecture of SDN controllers makes them susceptible to Distributed Denial of Service (DDoS) attacks, impacting network stability and service availability.

Purpose of the Study:

• To develop and evaluate advanced machine learning models for accurate and efficient detection of DDoS attacks in SDN environments.
• To enhance the precision and reduce false positives in DDoS attack detection through feature selection and hyperparameter optimization.

Main Methods:

• Implementation of Multilayer Perceptron (MLP) and Convolutional Neural Networks (CNN) for DDoS attack detection.
• Utilization of SHAP (SHapley Additive exPlanations) for feature selection to identify critical attack indicators.
• Application of Bayesian optimization for fine-tuning model hyperparameters to maximize performance.

Main Results:

• The proposed MLP and CNN models demonstrated exceptional accuracy in detecting DDoS attacks.
• Achieved a true positive rate of 99.95% on the CICDDoS-2019 dataset and 99.98% on the InSDN dataset.
• The SHAP feature selection and Bayesian optimization significantly improved detection efficiency and reduced false positives.

Conclusions:

• The integration of MLP and CNN with advanced optimization techniques provides a robust solution for DDoS attack detection in SDN.
• The findings highlight the effectiveness of machine learning in securing complex network infrastructures against sophisticated cyber threats.
• This research contributes a highly accurate and efficient method for mitigating DDoS attacks in Software-Defined Networks.