Intrusion Detection System Using ANN

Abstract

This paper proposes an advanced Intrusion Detection System (IDS) for IoT-based smart cities, utilizing Artificial Neural Networks (ANN) to enhance the detection of network anomalies with a 99% accuracy rate. Compared to CNN and LSTM-based models, this system introduces a multi-classifier capable of identifying five key network attacks: Denial of Service (DoS), User to Root (U2R), Remote to Local (R2L), Probe, and Other attacks. The IDS integrates with a user-friendly web application for real-time anomaly detection, attack type identification, and actionable preventive measures. The proposed model's superiority is demonstrated on the benchmark KDD Cup 1999 dataset, achieving significant advancements in classification accuracy and response time. This work contributes to securing IoT ecosystems by offering a scalable and reliable solution for smart city cybersecurity.