Intrusion Detection in Smart Cities: Leveraging Long Short-Term Memory Networks for Enhanced Cybersecurity

The use of the internet is continuously expanding, and data has become a crucial asset in today's digital landscape. Smart Cities leverage technology to develop innovative solutions for addressing urban challenges, utilizing communication and information frameworks. However, Smart City applications are susceptible to various cyberattacks, such as Probing (Prob), Denial of Service (DOS), User to Root (U2R) and Remote to Local (R2L) attacks. Prominent examples for Smart Cities include New York City, Dubai, London, Barcelona, Singapore, and Tokyo. This study introduces a deep learning-based way for attacks detection in Smart Cities by Long Short-Term Memory (LSTM) networks. The NSL-KDD dataset is utilized to assess the suggested solution, while the Synthetic Minority Over-sampling Technique (SMOTE) is activated to stem class imbalance. Experimental findings show that the model delivers high-performance metrics, achieving an accuracy of over 98.48%, a precision of 99%, a recall of 99%, and an F1-score of 98% on normalized data utilizing the SMOTE technique. These results focal point the validity of the proposed model in identifying attack traffic in Smart City environments.