Parallelism of AES Algorithm using CUDA-aware-MPI

I am working on a paper about paralleling AES algorithm using CUDA and MP{I together. Main idea is to improve performance according to using CUDA or MPI alone.

I have submitted my paper today (2014-03-09) to ISPA14 conference. You may see the abstract below;

Today, all around the world, people are striking or rioting for their privacy and communication freedom. Governments, Internet service providers or some third party organizations keep interrupting people’s communications [1][2][3][4]. The need of encryption, data and information privacy keeps increasing [5][6]. The US Government accepted AES as their national standard encryption algorithm to protect their own sensitive data [7][8]. Cryptography is a handful tool to secure data, to protect its integrity and to authenticate its origin [9]. Yet, encryption algorithms are based on complex mathematical techniques and they require computing power. Large data, like over 1GB, requires so much time to be encrypted. In some cases, required time can be higher than 10 minutes [6][10]. People always require data and they want their data to be protected. Most of these data should be processed real time (or very close to real time). All around the world, people buy their needs through online markets and online money transactions are a part of our lives. These transactions are very sensitive to keep persons identity safe, they shouldn't be interrupted, changed and they should be authenticated. Also, this has to be done in a short time. Serial approaches to encryption is not fast enough. So, in order to speed up, parallelism is a good way to start. This study is focusing on parallelism of AES algorithm using CUDA-Aware-MPI approach and comparing performance differences with CUDA and serial implementations. This study also one of the first studies which focuses on CUDA-Aware MPI approach.

by zgrw on 2014-03-12 13:45:08