SOFTWARE DEFINED NETWORK-BASED INTRUSION DETECTION IN CLOUD ENVIRONMENT USING MACHINE LEARNING

Abstract

Abstract-The continued adoption of cloud services has led to a surge in demand for more secure cloud environments — while our traditional Intrusion Detection Systems (IDS) are simply not cutting it when it comes to addressing the inherently multi-machine nature and seriously elastic business growth potential that modern day Cloud computing infrastructures enable. This feature proposes a novel technique by integrating machine learning (ML) with Software-Deined Networking principles to establish an eicient and reliable IDS framework applicable in cloud environments. With SDN controllers as a central system for network administration, the system facilitates real-time capture and analysis of packet data across an entire cloud infrastructure. The models are based on ML, which is trained to identify patterns and better find anomalies or abnormalities that could some underlying intrusion/ attack Africa in case. The framework improves cloud security by calibrating network policies on the fly and responding in real time to detected threats as well providing total, live visibility of their entire set-up across any Cloud. Extensive performance evaluations show that the proposed approach substantially outperforms previous methods, and produces 99.44% detection accuracy as well as much better precision recall F-score results compared to baseline methods in a real-world case study. Our results demonstrate the ability of proposed framework to tackle complexity faced in cloud security, and provide scaleable solution protecting clouds from new age cyber threats. The work presented in this article conclusively shows how to apply machine learning on network security monitoring using SDN (Software Defined Network) technologies which, I believe is a major new research direction for future secure cloud architecture developments.