Student : NE Ndlovu
About the student
Curious by nature and driven by innovation, I am Nolwazi Ndlovu, a Software Engineer who thrives on problem-solving and collaboration. With a strong passion for security, my research involves developing robust, multi-model intrusion detection systems, like the SMART SHIELDS project, to counter evolving cyber threats. My ultimate goal is engineering stronger, more secure digital infrastructures.
About the Project
Distributed Denial of Service (DDoS) attacks remain a critical threat, causing costly downtime and reputational damage. Traditional rule-based intrusion detection systems (IDS) struggle with the scalability and complexity of modern threats. The SMART SHIELDS project addresses this by designing and deploying an advanced, machine learning-driven IDS that is fast, accurate, and visually interpretable. Key Contributions: Weighted Ensemble Model: A robust system combining three complementary ML models—Autoencoder + Random Forest (99.71% accuracy), Isolation Forest (98.51% accuracy), and Two-Stage Naive Bayes (96.7% accuracy). The ensemble increases detection robustness and minimizes false positives. Modern Dataset Validation: The system was trained and validated on a 1.6 million-row subset of the comprehensive CICDDoS2019 dataset. Real-Time Deployment: The final system is deployed on a Streamlit dashboard , demonstrating the ability to provide real-time classification, network activity logs, and actionable alerts in a practical user interface. SMART SHIELDS successfully bridges the gap between theoretical machine learning research and practical cybersecurity application, providing a next-generation defense against evolving DDoS threats.
