Detecting Anomalies in the Data Residing over the Cloud
Autor: | Deepali Arora, Kin Fun Li |
---|---|
Rok vydání: | 2017 |
Předmět: |
Cloud computing security
Computer science business.industry Vulnerability 020206 networking & telecommunications Denial-of-service attack Cloud computing 02 engineering and technology Computer security computer.software_genre ComputingMilieux_MANAGEMENTOFCOMPUTINGANDINFORMATIONSYSTEMS 0202 electrical engineering electronic engineering information engineering 020201 artificial intelligence & image processing Algorithm design Anomaly detection business Cluster analysis computer Hacker Computer network |
Zdroj: | AINA Workshops |
DOI: | 10.1109/waina.2017.124 |
Popis: | With more companies turning towards cloud computing for storage and processing of their data, the security of the cloud becomes essential. However, cloud computing is vulnerable to many security threats, including data leakages, compromised credentials, presence of unauthorized users or entities, execution of insecure applications or programming interfaces and APIs, shared technology vulnerabilities, account hacking, malicious insiders, and denial of service (DoS) attacks. In this paper, by using anomaly detection technique in conjunction with k-means clustering, we demonstrate how users can be classified into malicious and non-malicious categories based on the activities they carried out while accessing data residing over the cloud. Additionally, it is also shown that by using supervised learning algorithm like SVM, it is possible to further classify malicious users into internal and external adversaries. Our results show that based on user activities, it is possible to identify both internal threats caused by the current or ex-employees intentionally or unintentionally, and external threats carried out by adversaries unknown to an organization. These results demonstrate that machine learning algorithms offer a promising solution in terms of identifying malicious and non-malicious users within a cloud framework, in a fast and efficient manner. |
Databáze: | OpenAIRE |
Externí odkaz: |