ABSTRACT
This chapter focuses on three types of threats to privacy that result from sharing users' media data via the cloud: The fragility of local software and hardware on mobile clients; the eavesdropping on the communication links; the untrustable cloud service providers. It aims to overhaul cloud media sharing by letting users assure themselves that no one can eavesdrop or understand what they are watching, posting, and communicating. To meet the objective of designing a format-compliant, compression-independent, and correlation-preserving cipher, the authors propose chained approaches via chaotic mapping, image-based key whitening, and Latin square pattern-based substitution. To lower the computational cost and encryption delay, the authors propose to utilize the GPU Shader for parallel pixel processing. The authors integrate all the proposed approaches into a customized image filter for easy use without modifying the existing code base. Experimental results demonstrate a sufficient security level for cloud media data.
