Detection and mitigation of DDoS attacks based on multi-dimensional characteristics in SDN.

Area of Science:

• Computer Science
• Cybersecurity
• Network Engineering

Background:

• Traditional Distributed Denial of Service (DDoS) detection and mitigation in Software-Defined Networking (SDN) suffer from high computational overhead, feature underutilization, and excessive bandwidth consumption.
• Existing methods often lack the efficiency and accuracy required for real-time threat management in dynamic SDN environments.

Purpose of the Study:

• To propose and evaluate a novel two-stage detection and mitigation method for DDoS attacks in SDN environments.
• To address the limitations of traditional methods by leveraging multi-dimensional traffic characteristics for improved accuracy and efficiency.

Main Methods:

• A two-stage approach involving coarse-grained detection via traffic statistics analysis and fine-grained detection using a Multi-Dimensional Deep Convolutional Classifier (MDDCC).
• The MDDCC employs wavelet decomposition and convolutional neural networks to extract multi-dimensional features from traffic data.
• Attack source tracing and isolation are achieved by integrating graph theory with restrictive strategies.

Main Results:

• The proposed method achieves quick and accurate DDoS attack detection in SDN networks using minimal statistical information.
• Experimental results demonstrate superior accuracy and generalization capabilities compared to traditional methods on both simulated and public datasets.
• Effective mitigation is achieved by isolating affected nodes, ensuring the continued transmission of legitimate traffic during attacks.

Conclusions:

• The developed method significantly enhances DDoS attack detection and mitigation capabilities in SDN environments.
• It offers a robust mechanism for containing cyber threats and safeguarding network integrity and performance.
• The approach provides a more efficient and accurate alternative to conventional DDoS defense strategies in SDN.