PRACTICAL POST-QUANTUM CRYPTOGRAPHY & FULLY HOMOMORPHIC ENCRYPTION ON GRAPHICS PROCESSORS: DESIGN & PERFORMANCE EVALUATION

Abstract

Lattice cryptography has seen rapid development in the past decade due to its role in enabling various emerging technologies such as post-quantum public-key cryptosystems and Fully Homomorphic Encryption (FHE) schemes that allow computing on encrypted data. In spite of the attractiveness of these technologies, they face a number of major challenges that slow down their adoption and use in practice. Essentially, these cryptosystems deal with large mathematical structures and require enormous computing power. This thesis tries to mitigate this problem by accelerating the performance of several lattice cryptographic schemes using Graphics Processing Units (GPUs). We propose both algorithmic and architectural solutions to design and implement a set of modern lattice-based cryptosystems taking into account the constraints imposed by the GPUs. The thesis also includes detailed performance evaluation and analysis to show the feasibility, benefits and limitations of current GPU platforms as a vehicle for accelerating lattice-based cryptography.