The Encrypted Block Cipher Algorithm Based On Gpu Parallel Clustering Research And Implementation

With the rapid development of the Internet of Things technology, the amount of data in the database increase whip and spur daily, we are buried in the mass data. Data mining is used to find decision-making information that enterprises need from the massive amounts of data, information security is used to protect the security of corporate decision-making information. K-Means cluster partitioning algorithm and AES encryption algorithms are traditional algorithm in data mining and information security field, but K-Means algorithm is computation huge and convergence of the centroid slow, AES algorithm has cumbersome calculation and many iterative times and so on. In recent years, GPU computing opens the door of parallel computing, GPU is ideal for floating-point operations that require high computational efficiency and low cost. CUDA (Compute Unified Architecture) introduced by Nvidia as a bridge between CPU and GPU, make GPU parallel computing no longer tedious.This article uses GPU-accelerated to combine K-Means clustering algorithm and AES block encryption algorithm and gives the an implement method of K-Means algorithm and AES block cipher algorithm based on CUDA. First this paper makes traditional K-Means clustering partitioning algorithm parallel improvements, enables it to meet the GPU operating condition, uses K-Means clustering algorithm based on CUDA to group the massive data object, and makes AES cryptographic algorithm parallel improvements,improves the Cipher Block model to make it could be executed concurrently in GPU, and then encrypts the results of grouping. This article compares and analyzes the performance of parallel GPU implementation results with parallel CPU.Finally, the paper has multiple sets of experimental tests about the algorithm,the effect of improved K-Means plus AES parallel algorithm based on GPU is better than the CPU parallel algorithm. Finally, the article compares its own algorithm with similar algorithms and finds that the its speed-up ratio is more obvious. Relative to the multi-core CPU architecture,the method of this paper which is GPU parallel algorithm can provide decision-making information to company quickly at the same time reduce the burden on company expenses, and then protect the decision-making information.