CLOUD INTRUSION DETECTION SYSTEM

Abstract: The rapid adoption of cloud computing has significantly expanded the cybersecurity attack surface, exposing infrastructures to volumetric network attacks and zero-day exploits. Traditional signature-based Intrusion Detection Systems fail to detect novel threats and generate excessive false alarms, while heavy deep learning models introduce severe computational latency. This paper proposes a lightweight, real-time Cloud Intrusion Detection System, utilizing the unsupervised Isolation Forest machine learning algorithm. By continuously analyzing network telemetry such as packet transmission rates and failed logins, the model autonomously establishes a baseline of normal behavior and mathematically isolates statistical anomalies. The algorithmic backend is seamlessly integrated into a custom-built, interactive Security Operations Center (SOC) dashboard using the Streamlit framework, providing live 3D threat vector visualizations and automated PDF auditing. Experimental results confirm the system effectively intercepts simulated brute-force and DDoS payloads with minimal processing overhead, establishing a highly proactive cloud defense mechanism.

Keywords: Cloud Security, Intrusion Detection System, Isolation Forest, Anomaly Detection, Streamlit, SOC Dashboard, Unsupervised Learning.