Research On Parallelization Techniques Of Algebraic Cryptanalysis Based On Heterogeneous Platforms

As an important supporting technology of network information security, cryptanalysis has played a more decisive role and aroused extensive concern nowadays. As one of the most widely used and fastest developed cryptanalysis techniques, algebraic cryptanalysis has achieved many breakthroughs. However, with the deepening of research,algebraic cryptanalysis is facing the challenge of high computational complexity of solving the algebraic system.To solve this problem, this paper studies the acceleration techniques of algebraic cryptanalysis, combining parallel computing technology, and has achieved the following research results:1.In order to implement parallel solving algebraic system on a heterogeneous platform effectively, the parallel computing model for solving the algebraic system is proposed by this paper based on the study on characteristics of existing parallel models and solving algebraic system. It provides the basic framework for parallel implementation of solving algebraic system on a heterogeneous platform.2.For taking advantage of coprocessor to accelerate solving algebraic system efficiently, this paper designs and implements the accelerating techniques of coprocessor by supporting matrix computation over F2, aiming at the existing algorithm.3.To improve the efficiency on solving algebraic system of coprocessor by taking full advantage of its computing resources, the method of performance evaluation and optimization for coprocessor is proposed based on the Roofline model.4.In this paper, the parallel implementation of Wu's characteristic set algorithm is completed on the CPU/GPU heterogeneous platform, and it is applied to algebraic cryptanalysis process for testing. By analysing the test results, the effectiveness of the proposed techniques in this paper for accelerating solving the algebraic system is verified.