AI-Powered Network Traffic Analysis: Simulation, Anomaly Detection and Classification

With the increasing complexity of network traffic, advanced methods for monitoring, analysis, and safeguarding network environments have become essential. This paper represents a robust framework for simulating, analyzing, and modeling network traffic to detect malicious anomalies, classify traffic, and identify potential security threats. A synthetic dataset mimicking real-world traffic was generated with key features such as timestamps, IP addresses, protocols, and packet sizes. Exploratory Data Analysis (EDA) revealed packet size trends and traffic behaviors over time. For anomaly detection, DBSCAN was employed, while traffic classification utilized Logistic Regression, Random Forest, and Long Short-Term Memory (LSTM) networks. All models demonstrated flawless performance, achieving 100% accuracy, precision, recall, and F1-score in both training and testing phases. LSTM-based time-series analysis also successfully captured patterns indicative of Distributed Denial-of-Service (DDoS) attacks. These results highlight the exceptional efficacy of the proposed machine learning approaches in accurately identifying irregular traffic and enhancing proactive network.