A Graphical Approach for Botnet Detection in IoT Edge Environments Using a Lightweight Dynamic Louvain Method

Abstract

The increasing adoption of Internet of Things (IoT) devices has increased the risk of botnet attacks, posing significant threats to device integrity, network performance, and user privacy. Existing botnet detection methods rely on computationally intensive network flow analysis, which is not suitable for resource-constrained IoT edge environments. This study introduces a novel graphical approach for botnet detection using a lightweight dynamic Louvain method. The method dynamically constructs temporal network graphs where nodes represent devices and edges capture the interactions. The graph topological features are extracted, and edge weights are integrated based on communication patterns. The communities are identified in the network by applying the dynamic Louvain method, and the anomalies in the community structure are analyzed to detect botnet activities. Experimental evaluations on the BoT-IoT dataset show that the proposed approach achieves 99.3% accuracy, 99.1% precision, 99.1% recall, and a 99.3% F1-score. Further, the proposed method is compared with the traditional graph-based approaches and demonstrates superior performance in terms of detection speed, scalability, and resource efficiency.