Enhancing Domain Generation Algorithm Detection in Botnet Environments Using Hybrid Deep Learning

View at publisher Summarise with VIRAT
Authors: Patchala Nageswara Rao, K. Venkata Ramaiah
Cited by: 0
References: 13

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

Domain Generation Algorithms (DGAs) significantly endanger cybersecurity since they allow malware and botnets to form a large number of domain names dynamically, making it difficult to detect and preventing its operation in the Internet of Things (IoT) and botnet context. Conventional machine learning methods, such as Support Vector Machines (SVM) and XGBoost, have demonstrated a positive outcome in detecting DGA in the context of botnet attacks, but fail to overcome the dynamic nature of contemporary Internet of Things (IoT) cyber threats. This paper provides a critical comparative analysis of conventional machine learning prototypes with a proposed hybrid Convolutional Neural Network- Recurrent Neural Network (CNN-RNN) architecture to support better botnet DGA detection using IoT setup. CNN element is used to extract spatial information of domain name characters sequence typical of command-and-control messages in botnet structures, and the RNN layer is an element that implicates time-delays and sequence patterns typical of DGA-created domains. We systematically compared the XGBoost, SVM, and proposed CNN-RNN hybrid model in terms of the accuracy, precision, recall, and F1-score measures using a large dataset of legitimate domains and DGA-generated malicious domains of different botnet families, including but not limited to the IoT-based domain targets. The experimental findings indicate that the adopted CNN-RNN hybrid architecture performs better with an accuracy of 98.7 and achieves a higher performance with the lowest possible false positive rates that are vital to the Internet of Things security models. The fact that the hybrid model can detect both local character-level trends and global sequential trends, renders it especially useful in detecting new and advanced forms of DGA, which are currently in use in the contemporary botnet infrastructure. These results can be used to make effective and adaptive intrusion detection systems that can protect the critical infrastructure against emerging botnet threats in Industrial Internet of Things (IIoT) systems.

Fields of Study & Subjects

Network Security and Intrusion Detection Spam and Phishing Detection Advanced Malware Detection Techniques

Publication details

Journal / source
International Conference on Communication, Computing and Electronics Systems (ICCCES)
Pages
1–9
Publication date
21 January 2026
Language
EN