Machine Learning Algorithms for Attack and Anomaly Detection in IoT

With the invention of IoT and its range of smart applications, people’s life has been reformed drastically. IoT infrastructure consists of actuators and sensors that generate a massive volume of data that requires extensive computing. Due to the constrained nature of IoT devices, they can be easily exploited. Moreover, some of the IoT nodes behaves abnormally which makes the IoT infrastructure vulnerable that could be exploited by the adversary to gain unauthorized access or to perform other malicious deed. Thus, anomaly detection is a primary concern in the IoT infrastructure, and the investigation of IoT network for detecting anomalies is a fast-emerging subject. There is a plethora of techniques which have been developed by the researchers for detecting anomalies. In this article, we have emphasized on the machine learning-based anomaly detection techniques due to their ability to bring out accurate results and predictions. Further, to provide a detailed overview, we have categorized the machine learning-based anomaly detection techniques into unsupervised and supervised learning. This study helps the researchers to get a better idea of machine learning techniques that have been employed for the detection of anomaly in the IoT infrastructure.