Machine Learning–Based IoT Botnet Detection: Techniques, Challenges, and Future Research Directions: A Comprehensive Review
Year: 2026
Type: article
Abstract: The rapid proliferation of the Internet of Things (IoT) devices has increased the attack surface of current networks, thus enabling IoT environments to be more penetrable to botnet-based cyber-attacks such as DDoS, scanning, malware spread, etc. The conventional simple security measures are often not much help because of IoT being highly diverse, highly resource-constrained, and large scaled. Therefore, ML and DL have been touted as promising mechanisms for facilitative-botnet detection in IoT. This review paper informs on those aspects of the recent ML methods from supervised, unsupervised, semi-supervised, and hybrid models that have been made use of in the exploration of novel ways of doing IoT botnet detections and analyses where forest Random, Support Vector Machine, ensemble, CNN, RNN, and Planning Systems are discussed critically in terms of detection efficiency, false alarm rate, and computational complexity. The paper also explores new paradigms like explainable artificial learning, federated learning, and the integration of cyber threat intelligence to promote the addictiveness and resilience of IoT systems. Even after achieving true detection accuracies greater than 95% from research exercises using benchmarked datasets, present approaches struggle with multiple shortcomings, which include class imbalance, lack of real-time deployment, high computational cost, nonexistent generalization capabilities to zero-day threats, and the fact that they do not work well for IoT devices. Through an extensive comparative analysis, a discussion of the existing vulnerabilities, and a summary of research gaps, this study altogether indicates the more desirable route of building efficient, scalable, and adaptive IoT botnet detection frameworks. While stressing the design of lightweight models, real-time detection, generalization across multiple datasets, and coupled prevention and detection mechanisms toward the building of resilient IoT cyber defenses, this paper discusses future research directions.
Language: English
Subfield: Computer Networks and Communications
Field: Computer Science
Domain: Physical Sciences
Open Access status: hybrid
OpenAlex ID: W7218330364