Australian Information Security Management Conference
Document Type
Conference Proceeding
Abstract
One of the challenges encryption faces is it is computationally intensive and therefore slow, it is vital to find faster methods to accelerate modern encryption algorithms to keep performance high whilst also preserving information security. Users often do not want to wait for applications to become responsive, applications on limited devices such as mobiles often compromise security in order to keep execution times quick. Often they use algorithms and key sizes which are not considered cryptographically secure in order to maintain a smooth user experience. Emerging approaches have begun using a devices Graphics Processing Unit (GPU) to offload some of the computational burden from the Central Processing Unit (CPU) in an effort to parallelize and accelerate the encryption algorithms. Programming for a GPU often involves the use of CUDA or OpenCL programming, however these approaches are platform dependant. This research focuses on utilizing a GPU to perform in-browser cryptography using WebGL and JavaScript. This allows any GPU-enabled device capable of launching an OpenGL compatible browser to perform GPU accelerated cryptography. A GPU based implementation of the NTRUEncrypt algorithm was created and tested against a CPU based version on a range of hardware devices with results, challenges and limitations discussed.
DOI
10.4225/75/58a6a626b43ed
Comments
Win, D., Hall, S., & Nisbet, A. (2016). Accelerating NTRUEncrypt for in-browser cryptography utilising graphical processing units and WebGL. In Johnstone, M. (Ed.). (2016). The Proceedings of 14th Australian Information Security Management Conference, 5-6 December, 2016, Edith Cowan University, Perth, Western Australia. (pp.60-66).