Font Size: a A A

Research And Implementation Of Cryptanalysis Software Framework Using A Genetic Algorithm

Posted on:2022-08-28Degree:MasterType:Thesis
Country:ChinaCandidate:X YangFull Text:PDF
GTID:2518306731497934Subject:Computer technology
Abstract/Summary:PDF Full Text Request
In today’s world,information is everywhere.The security of this information during storage and transmission is of utmost importance because it may be compromised,leading to economic loss,leakage of military or commercial secrets,and even life threatening.One of the widely used information security methods is encryption technology.Cryptanalysis is a process that attempts to undermine security.Under the condition of ciphertext only attack,the cryptanalysis problem can be regarded as a search problem,where the search space is all possible key combinations.Brute force attacks require a lot of resources and time,linear and differential cryptanalysis methods need to process a large amount of plaintext,and the use of heuristic methods such as Genetic Algorithm(GA for short)can reduce the search space to a considerable extent.The purpose of this paper is to combine the GA with the key search in cryptanalysis,improve the traditional GA selection operator,propose a new individual fitness evaluation technology,and design and implement a cryptanalysis software framework based on GA,improve the efficiency of cryptanalysis using GA.The main work of this paper is as follows:1.Aiming at the slow convergence speed and low efficiency of the traditional GA selection operator,two improved selection operators are proposed.The selection strategy is aimed at which chromosomes in the current generation will be used to reproduce offspring,and it is hoped that the next generation will have a higher degree of fitness.By analyzing the theory of traditional basic selection operators and their advantages and disadvantages,two selection operators that can produce better convergence speed are proposed.Experiments show that the truncation-based roulette selection has nearly doubled the convergence speed of the traditional roulette selection,and the deciphering time is reduced by 46.6%.The truncation-based competition selection is 1.3 times faster than the traditional competition selection,and the deciphering time is shortened 50.8%,the improved selection operator can increase the convergence speed of the GA,shorten the deciphering time,and greatly improve the efficiency of cryptanalysis based on the GA.2.Aiming at the single problem of GA’s individual fitness evaluation technology,an individual fitness evaluation model based on convolutional neural network is proposed.GA need a measure of individual fitness,a method of comparing partially decrypted ciphertexts with other partially decrypted ciphertexts.By analyzing the application principles of Convolutional Neural Networks in the field of text classification,a method based on Convolutional Neural Networks is proposed.The fitness evaluation technology realizes the evaluation of keys containing numbers and special characters,which effectively makes up for the deficiencies of the statistical feature method based on the frequency of letters.Experiments show that in the condition of the ciphertext length 1000 bytes,when the key length is 40 and 50 bytes,the decryption success rate can reach 100%,and when the key length is 60,70,and.80 bytes,the average deciphering success rates were 98.89%,98.57%,and 94.58%respectively.The individual fitness evaluation model based on Convolutional Neural Network is a very effective method for individual quality evaluation.3.Aiming at the problem of lack of a mature and easy-to-use tool platform for cryptanalysis research based on GA,a framework model of cryptanalysis software based on GA is proposed.In this paper,we integrate the encoding methods of different cryptosystems,realize the analysis and research on the codes of different cryptosystems,transplant the selection,crossover,mutation and other genetic operators of GA from other application fields to enrich the technical methods of genetic operators,and realize the scalability of genetic operators and the parameterized configuration of multiple genetic operators.Thus,we can reduce time-consuming and laborintensive basic work and save time and labor costs when using GA to carry out cryptanalysis problems.Experiments show that the proposed cryptanalysis software framework based on GA can better meet the functional requirements and realize the use of GA to carry out cryptanalysis and research work of different cryptosystems.
Keywords/Search Tags:Genetic Algorithm, Cryptanalysis, Selection Operator, Fitness Function, Convolutional Neural Network
PDF Full Text Request
Related items