Hybrid Intrusion Detection System For Securing Information In The Cloud Environment

Abstract

Cloud environments face escalating security risks due to their distributed and multi-tenant nature, rendering traditional intrusion detection systems insufficient. Cloud setting become attractive targets for various cyber-attacks besides these advantages, thus security is a big concern for both cloud consumers and providers. Traditional security standards may not work for dynamic, multi-tenant, distributed cloud architecture, necessitating more advanced and powerful intrusion detection solutions. Thus, we are proposing a Hybrid Intrusion Detection Model (HIDM) that integrates rule-based filtering, machine learning, and deep learning to enhance cloud network protection. The framework employs a three-layer detection strategy and ensemble weighted voting to minimize false positives while detecting both known and zero-day attacks. Evaluations are done using five benchmark datasets which includes NSL-KDD, CICIDS2018, UNSW-NB15, CIC-DDoS2019, and CIC Bell DNS EXF2021which shows that HIDM achieves 97.74% accuracy, 97.99% F1-score, and a 1.14% false positive rate, outperforming single-model approaches. There is various comparative analysis of accuracy are also given in the simulation section.  Unlike prior works, HIDM uniquely combines adaptive learning with real-time deployment capabilities, demonstrating a novel, scalable, and explainable approach to intelligent intrusion detection in the cloud.