IOT BOTNET DETECTION USING ARTIFICIAL INTELLIGENCE

Abstract

Recent findings draw the important security concern of botnets to the IoT devices and intensified by the surge in the number of connected devices and the swift development of advanced botnet networks. This paper is dedicated to the implementation of machine learning and deep learning algorithms to detect and classify botnet attacks in the IoT setting based on data, including UNSW-NB15. The machine learning pipeline provided is a complete cycle which includes exploratory data analysis and preprocessing of data, training, testing and evaluation of the data with different Machine Learning Algorithms. It is important to note here that the SVM model showed a 99.06% accuracy when classifying network traffic data as being normal or malicious with a slightly lower F1 score of 95.52. This model is the best in striking the right balance between accuracy and recall, correctly recognizing true positives and true negatives and giving a false alarm rate of 0.93% percent to produce few false positives and the classification of benign activities as detrimental. This approach will help in detecting botnets attack proactively and provide a preemptive approach to prevent future attacks. The system, based on machine learning methods, is effective at identifying and classifying botnet attacks, and is scalable, and can scan more than one IoT object at a time, even botnet threats it hadn’t known before. Its convenient design allows it to integrate with the existing IoT devices with ease. Overall, the given suggested solution is a solid way of identifying botnet attacks and mitigating them in the IoT environment. Using the strength of machine learning, the system offers scalable and efficient detection functionality and attempts to protect IoT devices against possible botnet threats and improve the overall security levels.