Detection and Prevention of Cyber Attacks on Cloud-Based Data Centers using Machine Learning

Abstract

The security issues with cloud computing have seen very high growth in recent times. This growth is certainly due to the increase in usage and adaptation of cloud-based services. The benefits from the cloud-based services or the data centers have improved the application building and application management life cycle. Hence, in the last 10 years, significant growth in data and data-driven applications can be seen on the cloud. This has indirectly increased the possibility of attacks on the cloud data. The data on the cloud is primarily hosted on virtual machines or virtualized storage devices. These devices are remotely manageable; thus, the attackers can also manipulate the device securities remotely or can generate cyber-attacks. In recent times, a good number of parallel research outcomes can be seen aiming to reduce the security issues on cloud services. Nonetheless, these parallel research outcomes are highly criticized for a variety of reasons. Firstly, these methods are highly time complex to detect any attack. Thus, cannot be considered reactive attacks, which is the most desirable type of security solution due to less complexity. Secondly, these methods are not applicable to detect a newer type of attack on the cloud or distributed network architectures. Henceforth, this work aims to solve these existing challenges using machine learning methods. This work proposes a novel algorithm to reduce the size of the network information without losing the critical information for detecting the attacks. This ensures timely detection of the attacks and can be preserved as reactive security protocols. Further, this work deploys another novel algorithm for detecting the anomalies based on the network connectivity characteristics using the rule mining methods. This will ensure the detection of the newer types of attacks also based on the deviation of the network characteristics. The proposed algorithms demonstrate nearly 99% accuracy in detecting cyber-attacks on cloud-based services.