AI-Driven Detection of Network Traffic Anomalies: A Case Study with OMNeT++

This study explores <b>machine learning-based anomaly detection</b> for <b>Distributed Denial-of-Service (DDoS) attacks</b> using the <b>OMNeT++ simulation platform</b>. Network traffic features such as <b>packet size, ports, and transmission frequency</b> were collected to train <b>Logistic Regression, Decision Trees, MLP, and Random Forest</b> models.The <b>Random Forest model achieved the highest accuracy (94%) and ROC-AUC score (0.927)</b>, demonstrating its effectiveness in intrusion detection. The study emphasizes <b>data preprocessing, feature selection, and class balancing</b> for improved model reliability.Future work will focus on <b>real-time detection, adaptive learning, and expanded simulations</b> to enhance network security