Behavioral Profiling for Transparent Verification in Cloud Storage Services
Security is still the most sensitive issue in cloud storage services as services remain accessible to users for prolonged periods following an initial (usually simple) authentication. ThisAl-Bayati B, Clarke NL, Haskell-Dowland PS (Dowland PS), Li F
issue has led to an increased vulnerability to potential attacks and sensitive customer information
being misused. To that end, this paper investigates behavioural profiling for continuous and transparent authentication and assesses the legitimacy of users while they interact with an example
cloud service—Dropbox. The best experimental result achieved an EER of 5.8%, based on an
implementation using a real-life dataset. The results are encouraging and indicate the feasibility
of detecting misuse in Dropbox.