Securing Cyberspace: Traditional vs. IoT Botnets-A Machine Learning Classification Approach
Sajal Gupta, John Alexander, Kanchan Bartwal, Harshit Narang, Daksh Rawat, Manisha Aeri · 2024
In the modern technology era where almost all devices are connected by IOT and are increasing exponentially every day, the threat of cyber- attacks against such devices is also increasing. Taking all the possible problems the devices face into consideration while being connected to the internet poses a serious threat to the security of these devices. A big problem is the creation of IoT botnets. An IoT botnet is a collection of infected IoT devices that cybercriminals can control from a distance for malicious purposes. These botnets are mostly used to start various cyber-attacks including the DDoS attacks, data theft, ransomware attacks, etc., which can cause serious damage to anyone including individuals, organizations, and even the country. Hence, the problem referred to by this research paper is to develop a robust and accurate method to detect and classify IoT botnets on devices. The methodology mentioned in this research paper is helpful in detecting different types of botnets and classifying them into various groups which are based on their behavior, network characteristics, and attack patterns. It can also classify them into the following two compromised and uncompromised devices. Moreover, the method should be scalable and efficient to handle a good and large number of IoT devices and real-time data streams. The final result after comparing the performance of different models gives the performance of Light Gradient Boosting Machine Classifier as the best with the accuracy of 99.85 in Validation Set and a minimal latency of 4 seconds and 99.86 in Test Set and a minimal latency of 5.2 seconds.