Integrated Effecient Approach to Botnet Detection using Supervised Machine Learning

Abstract Botnets are one of the most significant threats to cybersecurity. These are networks of compromised nodes of computers and other devices on the internet commonly used for a wide range of malicious activities across the globe. When deployed they can cause immense damage to computer systems by infecting them with malware in the likes of worms, trojans, viruses, rootkits, spyware, etc., stealing data, and launching large-scale attacks. As botnets are becoming increasingly sophisticated, traditional methods for detecting them are no longer effective. In this research paper, we propose a fully distributed collaborative botnet detection model using a binary classifier that decomposes the problem based on the expertise and experience of a distributed collaborative network, to efficiently detect a wide range of botnets. We measured the accuracy of collaborators’ responses ta o series of requests that we broadcast over a period of time accurately and efficiently detect botnets. We used 4 different classification algorithms to process the dataset. We evaluate our approach on a real-world dataset and each classifier outperforms existing methods for botnet detection and performed well with an average detection rate of 93.5%. Effective algorithms were also developed to improve the reputation and robustness of the detection system.