A Systematic Approach for IoT Cyber-Attacks Detection in Smart Cities Using Machine Learning Techniques

Abstract

In these last years, the widespread adoption of the Internet of Things (IoT) concept led to the invention of intelligent cities. Smart cities operate in real time to promote lightness and life quality to citizens in urban cities. Smart city network traffic through IoT systems is growing exponentially though it presents new cyber-security threats. To deal with cyber-security in smart cities, developers need to improve new methods and approaches for detecting infected IoT devices and cyber-attacks. In this paper, we address IoT cyber security challenges, threats and solutions in intelligent cities. We propose an approach for anomaly detection in smart cities applications, networks and systems. Our solution relies on intelligent anomalies as vulnerabilities and threats detection based on different methods and machine learning algorithms. The proposed solution helps in effectively detecting and localizing infected IoT devices as well as generating alerts and reports. To experiment our solution, we used the dataset NSL-KDD to evaluate the accuracy of the model. Obtained results show that our model achieved a high classification accuracy of 99.31% with a low false positive rate.