IoT BotScan Ultra-Lightweight AI Defense Against Botnet Threats.

The rapid growth of IoT environments has increased exposure to botnet-based attacks that disrupt system behavior and consume network resources. Traditional detection methods often require heavy computation and complex deployment, making them unsuitable for lightweight environments. This project, titled “IoT BotScan: Ultra-Lightweight AI Defense Against Botnet Threats”, presents an efficient detection framework using machine learning and deep learning techniques. The system analyzes network traffic patterns from the N-BaIoT dataset to identify malicious activities. Multiple algorithms such as Random Forest, XGBoost, LightGBM, Convolutional Neural Networks, and a stacking-based ensemble model are implemented to improve detection accuracy and stability. The proposed framework is designed using a Flask-based web application that provides secure access through authentication modules and a prediction interface. Feature extraction and model training are optimized to reduce complexity while maintaining reliable performance. The stacking approach combines the strengths of individual models to enhance classification capability. This project demonstrates that lightweight AI models can effectively detect botnet behavior without excessive processing overhead. The system aims to support secure IoT communication by providing a scalable and efficient defense mechanism suitable for research and academic analysis.

Keywords: IoT Security, Botnet Detection, Machine Learning, Deep Learning, Random Forest, XGBoost, LightGBM, CNN, Stacking Model, Flask.