Intelligent Cyber Attack Detection through Machine Learning

Abstract

Internet of Things (IoT) Connection encouraged new technology solutions for public safety in the city by integrating the transportation, healthcare and environment sectors into smart services. The variety of devices and continual chug of information makes these interdependent systems effective for both improving operational effectiveness and enhancing quality of life, but provides cybercriminals with a broad attack surface. Existing methods of security are simply not up to the task of addressing realtime, complex threats. In this work, we present a novel machine learning-based approach to detect a cyber attack in Internet of Things (IoT) powered smart cities. This proposed solution would be based on the analysis of network traffic and activities of the devices to determine whether the activity is normal or not. In the data preparation section, unimportant features are removed, missing values are filled, scaling and filters of features are applied on the data to improve the accuracy of the model. Various approaches are implemented and explored to find the best machine learning approach for detecting cyberattacks. Generally, accuracy, precision, recall, F1 score and confusion matrix analysis are used to evaluate the performance. Its aim is to keep the computational efficiency, increase early Cyber intrusion detection and decrease false alarms. The patterns learnt may be finding an abnormal behaviour in historic activity that can be used to detect Denial of Service (DoS), probing, access or data manipulation attacks. The results of the experiment highlight an immense improvement in the time taken for the detection of threats, from nearly real-time time to almost instant time, with the use of machine learning threat detection technology, which greatly improves the security of the IoT infrastructure. The study results underline the importance of Smart City Security Systems. The proposed architecture is flexible and extendable with the number of urban services and in the future it will offer cyber resilience capabilities to the IoT based system.