An Improved Mechanism for Investigation, Detection, and Classification of Iot Botnet Attacks

Many industries and scientific societies are now interested in Internet of Things (IoT) technology because of its many intelligent uses. In particular, the quantity of suspicious activity such as IoT botnets and extensive cyberattacks, has sharply increased, as has the number of IoT devices that are vulnerable or unprotected. Numerous conventional techniques for detecting botnets struggle to scale up to the demands of secured IoT network. The conventional approach has scalability problems that impact all aspects of the Botnet detection system, including feature extraction, data collecting, storage, and analysis, and are not limited to detection bottlenecks. The main objective of the current study is to create a classification model that can identify different types of IoT botnet attacks. There are several ways to attack the IoT architecture's network and application layers. The proposed approach is to evaluate how safe and, more importantly, resilient commonplace IoT devices can be against the Mirai and Gafgyt botnets. In this article, we have crafted and implemented four different models such as random forest, autoencoder, linear, and threshold classification that is utilized to detect and classify the IoT botnet attacks. The validation has been done by using zero-day attack classification.