Securing IoT Using Supervised Machine Learning

IoT today is ubiquitous. Its growth has reached a stage where the number of connected devices is growing exponentially daily, raising various security issues of great importance to all stakeholders. Implementing security schemes on IoT devices poses a considerable challenge because of their heterogeneous and restricted existence. Machine learning is already an intricate part of several IoT applications, and it eliminates human errors and enables IoT infrastructure to generate real-time insights to reach its full potential. To protect IoT services from being targeted and raise security awareness throughout all aspects of the network, machine learning algorithms are now explored to address the critical security issues of IoT. Deploying a learned model-based security approach is dynamic and holistically effective as a solution. Supervised machine learning methods have been proposed as a promising approach for IoT security due to their ability to detect and classify malicious data. This paper provides a comprehensive overview of supervised machine learning methods used for IoT security, including various classifiers and data engineering techniques. We have further demonstrated how effectively the performance can be increased by deploying various data engineering methods to improve the overall model performance in terms of accuracy and time consumed. Both these play a crucial role in providing dynamic and real-time services. Overall, this paper aims to raise awareness of the potential of supervised machine learning for IoT security and guide researchers and practitioners in this field.