ANOMALY DETECTION IN IOT NETWORK TRAFFIC

Abstract

With the rapid growth of the Internet of Things (IoT), billions of interconnected devices communicate continuously, generating massive amounts of network traffic. While this connectivity offers convenience and automation, it also introduces significant security challenges. Due to limited computational resources, IoT devices often lack strong security mechanisms, making them easy targets for cyber attackers. Traditional intrusion detection systems are insufficient for identifying subtle or zero-day attacks in these networks. This project proposes an intelligent, lightweight, and accurate machine learning–based anomaly detection system to monitor IoT network traffic and identify abnormal behavior in real time. The system uses the Isolation Forest algorithm to distinguish between normal and malicious network patterns. Data preprocessing and normalization are performed using PyShark, Pandas, NumPy, and StandardScaler to ensure consistency and reliability. The system provides real- time alerts for detected anomalies and offers scalability for deployment across various IoT platforms. Experimental results show that the model can effectively detect anomalies while maintaining low computational overhead, thus ensuring security and performance for IoT environments.