Machine Learning–Based Security in Cloud Database—A Survey
Utsav Vora, Jayleena Mahato, Hrishav Dasgupta, Anand Kumar, Swarup Kr Ghosh · 2021
Cloud computing has grown as an innovative technology providing physical as well as logical resources over the internet and changing the way of collecting and sharing data files. This growth is a reflection of the rapid transition and intense interest of academic and industrial organizations in this technology. Despite having several advantages, cloud security has always been a concern for cloud services. With the evolution of technology worldwide, the threats are also getting more complicated and advanced such as data breaching, insecure interfaces, shared technology vulnerabilities, and distributed attacks. A safe environment for cloud services is highly essential. For this, Intrusion Detection Systems based on various Machine Learning (ML) algorithms, Supervised, Unsupervised, or Hybrid, have been developed. In this review, we focus on understanding the current status of ML algorithms and emphasize those Intrusion Detection Systems which are evaluated using either of the two worldwide recognized datasets, NSL-KDD dataset, and UNSW-NB15 dataset. Related research papers are examined with respective experimental setups and other important factors. We stress our research on addressing security threats, issues related to cloud security, and the advantages and disadvantages that are faced on implementation of ML for detection of attacks in the cloud. We also aim to deliver effective solutions and put forth some open challenges as well for future research and development on cloud security.