A CNN based deep learning model for detecting P2P Botnets using flow features
H. G. Mohan, Jalesh Kumar, I. S. Rajesh, K Sai Geethanjali, Manjunath Sargur Krishnamurthy, K Jagannathan · 2024
A botnet is a network of computers managed by a botmaster or a command and control (C&C) server. Botnets are a big concern on the cyber security. The Peer-to-peer (P2P) applications have made considerable advances in these days. The P2P networks are is widely used for file sharing and multimedia applications, because to its high transmission rate and robustness. The P2P botnets are typical examples of harmful malwares directed to execute different harmful operations. In this work, a novel Convolutional Neural Networks (CNNs) model is used to identify the P2P botnets using traffic flow features. Unlike traditional botnet detection methods, the CNN model incorporates the spatial feature representations, providing a comprehensive analysis of network behavior. The model is validated on BoT-IoT dataset to detect botnets along with its categories and subcategories of attacks. The model achieved 99% of accuracy and an AUC of 0.94. Through ROC plots it is evident that the model is robust in identifying the category and subcategory of attacks. The experimental results demonstrate that our CNN model significantly outperforms existing state-of-the-art methods in terms of detection accuracy and false positive rate.