The Implementation Of Parallel AES Encryption Algorithm Baesd On GPU

In recent years, the network technology develop rapidly, the demand forinformation security is also becoming increasingly high. Advanced EncryptionStandard(AES)encryption algorithm as a representative of cryptography today hasvery broad prospects for development. However, due to the impact of CPU serialstructure, the speed on the implementation of traditional serial AES encryptionalgorithm is not satisfactory；If you are using FPGA and other programmable logicdevice to implement the accelerated AES encryption devices,there will be a series ofproblems which increase costs and limit the scope of application , such as thedevelopment is difficult, hard to extensions, hardware upgrades need to modify thecode, etc. In this way, looking for how to improve the speed of AES encryptionalgorithm method has become a top priority.As a new calculation resource,GPU (Graphic Processing Unit) has a lot ofparallel processing framework advantage . Compared to the slow development ofCPU computing ability, in order to meet the computer simulation and computer gamesand other real-time requirements of market expectations,GPU 's floating pointcomputing ability and data bandwidth since 2003 has far exceeded the CPU .GPUsuites large-scale parallel computing .A lot of domestic and foreign researchers wereabstracted to the GPU encryption area, researching a way to use the free resourceefficiently. This shows that the research on speeding up the AES algorithm is of greatrealistic significance. There are many ways of GPU programming and CUDA(Compute Unified Device Architecture) is the most widely accepted one, which is the reason for the author to choose CUDA as the platform to realize the parallelism ofAES encryption algorithm. The main works of the paper are as follows:On the related knowledge of AES, this paper gives a brief introduction of themathematical basis of AES and a detailed introduction of the process of encryption.The module of Parallel AES Encryption Algorithm is summarized by analyzing thework mode of Parallel AES Encryption Algorithm. On the programming knowledgeof GPU, this paper introduced the main point of the usage of CUDA parallelprogramming platform, giving the realization of Parallel AES Encryption Algorithm atechnical support. On the improvement and realization of the algorithm, this papergive a improve method of the encryption round functions of AES, decreasing thenumbers of the execute steps. The analysis and realization are performed on CUDAplatform according to the very Parallel AES Encryption Algorithm. In the end part, atest and analyze between the original serial CPU AES algorithm and the Parallel AESEncryption Algorithm designed in this paper is implemented ,summarizing the maininfluence factor of Parallel AES Encryption Algorithm on GPU. An intuitional testresult is given out and a performance appraisal is made.The purpose of the paper is to increase the encryption speed of the AESalgorithm. The author uses the free resource of the GPU to carry out the design andrealization of the Parallel AES Encryption Algorithm, exerts the advantage of theparallel computing ability of GPU and makes a comparison with the traditional serialAES Encryption Algorithm to prove that the encryption speed of the Parallel AESEncryption Algorithm on GPU is ideal by the program test result, which providing anew thinking of speeding up the AES algorithm.