Abstract:

The Internet of Things (IoT) has enabled a massive expansion of interconnected devices, ranging from smart home appliances to industrial machines. However, this apid growth has also introduced new vulnerabilities, making IoT devices common targets for cyber-attacks, especially botnet attacks. An IoT botnet is a network of compromised devices remotely controlled by attackers. These devices are often used to carry out Distributed Denial of Service (DDoS) attacks and other malicious activities. Due to the diversity and resource limitations of IoT devices, detecting these botnets is particularly challenging. This project focuses exclusively on detecting IoT botnet attacks using machine learning (ML). It aims to create a model capable of analyzing network traffic and identifying malicious behavior by learning from patterns in data. Machine learning models are trained on labeled datasets containing both benign and malicious traffic. Features such as packet size, duration, frequency, and protocol types are extracted to help the model learn to distinguish between normal and suspicious activity. Various ML algorithms like Decision Trees, Random Forests, SVMs, and Neural Networks are evaluated based on metrics like accuracy, precision, and recall. The model with the best performance is selected for real-time detection. Data preprocessing, including normalization and encoding, is crucial to prepare the dataset for effective training. Datasets like Bot-IoT is commonly used for this purpose. The proposed detection system demonstrates high accuracy in identifying botnet traffic, providing a reliable tool for network monitoring and threat detection in IoT environments. In future work, deep learning and anomaly-based techniques may be integrated to improve detection of new and unknown botnet behaviors. Edge computing may also be explored for real-time deployment.

Description:

.