Authors: Hulwale Aditya Jaydas, Prof. C. A. Kolse, Prof. M. C. Kolse, Prof. S. K. Said
Abstract: The rapid growth of internet-connected computers and Internet of Things devices has expanded the pool of machines that attackers can compromise and organize into botnets. A botnet is a network of infected devices controlled by an attacker through command-and-control channels and used for distributed denial-of-service attacks, spam, credential theft and further intrusion. Signature-based tools struggle to detect new botnet families, encrypted communication and rapidly changing infrastructure. This paper reviews Machine Learning (ML) approaches for detecting botnet activity from network traffic. It covers botnet architectures and life cycle, traffic features derived from flows such as packet size, connection duration, destination diversity and communication periodicity, and detection methods ranging from clustering of correlated behaviour and supervised classifiers such as Random Forest, Decision Tree, Support Vector Machine and k-Nearest Neighbours to autoencoder-based anomaly detection for IoT devices and recurrent models for algorithmically generated domains. Benchmark datasets including CTU-13, N-BaIoT and Bot-IoT are discussed. A layered detection architecture with traffic collection, feature engineering, hybrid detection, correlation and response is proposed, together with research gaps, objectives and a comparison of algorithms. The review concludes that combining flow level ML with host correlation and continuous retraining provides early and reliable botnet detection.
