Neural Network Based Botnet Detection
Ritik Raj Singh, Kaunik Kamila, J. Kalaivani · 2021 International Conference on Intelligent Technologies (CONIT) · 2021
Botnets are one of the most recurring and severe threats for businesses and people, and they are difficult to identify using conventional approaches. In order to prevent detection by detection bots, botnet operators use a range of hiding methods, network topologies, and communication protocols. As a result, protecting against botnet attacks is an arduous challenge, and numerous studies for botnet identification have been suggested. Researchers have used machine learning approaches with various feature sets as a common methodology. Based on insights about botnet, a selected feature are used for machine learning algorithms for classifying botnet based traffic. These methods, conducted against botnet traces, had shown reasonable detection outcomes. However, their usefulness in the identification of different botnets or actual traffic remain uncertain. In this paper, we used different combinations of features to provide more information for detection that has not been extensively analyzed. In order to ensure proper assessment, we built a data set with a variety of botnet traces and background traffic. A proposed multilayer neural network training model is built with these features, and we studied their relative efficacy using this model.