Construction And Analysis Of S-Box Based On Intelligent Computing

As the only nonlinear component in block cipher algorithm,S-box directly affects the security of cryptographic system,so the construction of S-box with excellent cryptographic properties has always been one of the important research directions in the field of cryptography.Due to the complexity and variability of cryptographic attacks,it is difficult to construct S-boxes with "high" cryptographic strength by traditional methods.In recent years,the rapid development of intelligent computing provides a new idea for the construction of S boxes.Thesis applies intelligent computing to the construction of S-boxes with excellent cryptographic performance,and proposes directed mutation genetic algorithm and improved directed mutation genetic algorithm.The former of which can construct S-boxes with high nonlinearity,and the latter of which can construct S-boxes with not only high nonlinearity but also low differential uniformity and excellent balance.The research work of thesis is as follows:(1)A directed mutation genetic algorithm for constructing highly non-linear S-boxes is proposed based on crossover and mutation strategies.Firstly,the construction schemes of S-boxes based on traditional genetic algorithms are analyzed,and it is found that the correlation between the evolution direction established by these schemes based on fitness function and nonlinearity and the nonlinearity of S-boxes is weak,so it is impossible to quickly construct highly nonlinear S-boxes.Thesis proposes a directed mutation genetic algorithm,which optimizes two aspects based on the traditional genetic algorithm structure.Firstly,based on the relationship between nonlinearity and the maximum Walsh spectrum,a strong correlation between evolution direction and high nonlinearity S-boxes has been accurately established,enabling the designed directed mutation genetic algorithm to quickly construct high nonlinearity S-boxes;Secondly,according to the characteristic that S-box is the arrangement of multiple Boolean function,a strategy of crossing a single Boolean function is proposed to improve the probability of generating a balanced S-box.Through simulation verification,the probability of constructing a balanced S-box using directed mutation genetic algorithm is close to 50%.Compared to traditional algorithms,the constructed S-box has better nonlinearity indicators and a convergence speed of 40% faster.(2)Introducing chaotic mapping into directed mutation genetic algorithm to improve algorithm search efficiency.By analyzing the characteristics of chaotic systems,it was found that the complexity of chaotic value changes can establish a transformation relationship with the diversity of the initial population of genetic algorithms.A method of generating the initial S-box population using Logistic mapping instead of pseudo-random number generators was proposed.The simulation results show that the algorithm incorporating chaotic mapping improves its global search ability by nearly10%.(3)An improved directed mutation genetic algorithm is proposed by introducing exchange operation and balance adjustment for the two indicators of balance and differential uniformity.Firstly,based on the impact of input changes on the S-box differential matrix,the exchange operation is introduced to accurately establish the direct relationship between the evolution direction and the low differential uniformity S-box,enabling the improved directed mutation genetic algorithm to quickly construct low differential uniformity S-boxes;Secondly,according to the characteristics of the truth table of the balanced S-box,the balance adjustment operation is introduced,so that the S-box constructed by the improved algorithm is balanced.Through simulation verification,the probability of constructing a balanced S-box using the improved directed mutation genetic algorithm is 100%,and the nonlinearity and differential uniformity indicators of the constructed S-box are better than other algorithms.