Related Experiment Videos
LwHM: lightweight hybrid classifier for SDN-attack detection using recursive feature elimination
Khadija Kanwal1, Muhammad Mujahid2, Julio Cesar Martinez Espinosa3,4,5,6
1Institute of Computer Science and Information Technology, The Women University Multan, Multan, Pakistan.
Scientific Reports
|June 24, 2026
Summary
Artificial intelligence enhances software-defined networking (SDN) security by improving data integrity and using feature selection. A hybrid AI model achieved 99.93% accuracy in detecting network intrusions.
Area of Science:
- Cybersecurity
- Artificial Intelligence
- Network Security
Background:
- Increased reliance on internet connectivity necessitates robust cybersecurity measures.
- Cybersecurity threats to data, privacy, and critical systems are prevalent.
- Software-defined networking (SDN) requires advanced security solutions.
Purpose of the Study:
- To investigate the role of artificial intelligence (AI) in enhancing SDN security.
- To propose a comprehensive three-part approach for securing SDN environments.
- To develop and evaluate a lightweight hybrid model (LwHM) for intrusion detection.
Main Methods:
- Data integrity assurance through cleaning, preprocessing, and normalization of an SDN intrusion dataset.
- Application of six feature selection strategies: RFE, polynomial features, ANNs, SelectKBest, LASSO, and correlation-based features.
- Development of a lightweight hybrid model (LwHM) using k-nearest neighbors and decision trees with a voting classifier.
Main Results:
- Feature selection techniques identified significant features, improving model performance.
- The LwHM demonstrated superior performance on the InSDN dataset.
- An accuracy score of 99.93% was achieved using Recursive Feature Elimination (RFE) features.
Conclusions:
- The proposed AI-driven approach effectively enhances SDN security.
- The LwHM offers an efficient solution for thwarting network breaches.
- AI plays a crucial role in strengthening cybersecurity for modern networks.