Cyberattacks Detection in IoT-Based Smart City Applications Using Machine Learning Techniques

Area of Science:

• Cybersecurity
• Machine Learning
• Internet of Things (IoT)

Background:

• Smart cities leverage IoT for efficiency and quality of life, but face escalating cybersecurity risks.
• IoT devices in smart cities are vulnerable to malicious attacks due to network connectivity.
• Effective defense mechanisms are crucial to prevent IoT device failures and data breaches.

Purpose of the Study:

• To explore machine learning algorithms for detecting cyberattacks and anomalies in smart city IoT networks.
• To evaluate the performance of single classifiers versus ensemble methods for improved threat detection.
• To integrate feature selection, cross-validation, and multi-class classification for robust cybersecurity.

Main Methods:

• Utilized machine learning algorithms: Logistic Regression (LR), Support Vector Machines (SVM), Decision Trees (DT), Random Forests (RF), Artificial Neural Networks (ANN), and K-Nearest Neighbors (KNN).
• Implemented ensemble techniques: Bagging, Boosting, and Stacking to enhance detection system performance.
• Incorporated feature selection, cross-validation, and multi-class classification for comprehensive analysis.

Main Results:

• The proposed technique effectively identifies cyberattacks within smart city IoT environments.
• The stacking ensemble model demonstrated superior performance over other models.
• Stacking achieved higher accuracy, precision, recall, and F1-Score in attack detection.

Conclusions:

• Machine learning, especially ensemble methods like stacking, offers a promising solution for smart city IoT cybersecurity.
• The integrated approach of feature selection, cross-validation, and multi-class classification enhances detection capabilities.
• Further research into stacking ensembles can lead to more resilient smart city networks.