Cyber Resilience in Smart Cities Using Bi-LSTM-Based Intrusion Detection System

Cybersecurity risks have grown as smart technology is being integrated into urban infrastructure. Intrusion detection systems (IDS) defend networked systems from emerging threats, making them essential for smart city cyber resilience. Regular intrusion detection systems (IDS) have false alarm rates, delayed threat detection, and an inability to handle complex threats. Given that these vulnerabilities threaten smart city network safety and effectiveness, a robust real-time anomaly detection system is needed. This paper suggests a novel technique called Bi-Directional Long Short-Term Memory for Intrusion Detection Systems to build smart cities’ cyber resilience. To improve anomaly detection, the Bi-LSTM model considers network data temporal linkages. The proposed system is trained and verified using benchmark datasets, and feature selection and hyperparameter tuning increase performance. The simulation outcomes demonstrate that the suggested model increases the attack prediction accuracy ratio by 98.9%, computational efficiency ratio by 97.8%, and reduced end-to-end detection latency of 9.2%, average inference time of 15.4% compared to other existing models.