Substitution boxes (S-boxes) serve as critical nonlinear components in modern symmetric cryptography, directly influencing the security of block ciphers against various cryptanalytic attacks. The generation of cryptographically strong S-boxes with optimal nonlinearity represents a computationally challenging optimization problem, given the vast search space of 256! possible permutations for 8-bit bijective functions. Traditional genetic algorithms, while promising for such complex optimization tasks, suffer from poor convergence rates, excessive computational requirements, and low success rates when applied to S-box generation. This chapter presents a novel selection algorithm that addresses the fundamental limitations of standard genetic algorithms for S-box optimization. The proposed hybrid approach combines population-based exploration with local optimization principles, integrating the diversity-preserving capabilities of genetic algorithms with the exploitation efficiency of hill-climbing methods. The algorithm employs intelligent selection mechanisms that preserve high-quality individuals while promptly eliminating poor solutions, thereby significantly reducing computational overhead. Comprehensive experimental validation demonstrates substantial performance improvements over existing approaches. The algorithm achieves a 99% success rate in generating S-boxes with optimal nonlinearity of 104 https://www.w3.org/1998/Math/MathML" display="inline"> (N f = 104), requiring only 53,000 fitness evaluations on average. This corresponds to roughly a 50% reduction in computational complexity compared to standard genetic algorithms and enables practical real-time S-box generation with execution times as low as 8–9 seconds on modern multi-core processors.
Advanced Cryptographic Algorithms
Oleksandr Kuznetsov;
2026-01-01
Abstract
Substitution boxes (S-boxes) serve as critical nonlinear components in modern symmetric cryptography, directly influencing the security of block ciphers against various cryptanalytic attacks. The generation of cryptographically strong S-boxes with optimal nonlinearity represents a computationally challenging optimization problem, given the vast search space of 256! possible permutations for 8-bit bijective functions. Traditional genetic algorithms, while promising for such complex optimization tasks, suffer from poor convergence rates, excessive computational requirements, and low success rates when applied to S-box generation. This chapter presents a novel selection algorithm that addresses the fundamental limitations of standard genetic algorithms for S-box optimization. The proposed hybrid approach combines population-based exploration with local optimization principles, integrating the diversity-preserving capabilities of genetic algorithms with the exploitation efficiency of hill-climbing methods. The algorithm employs intelligent selection mechanisms that preserve high-quality individuals while promptly eliminating poor solutions, thereby significantly reducing computational overhead. Comprehensive experimental validation demonstrates substantial performance improvements over existing approaches. The algorithm achieves a 99% success rate in generating S-boxes with optimal nonlinearity of 104 https://www.w3.org/1998/Math/MathML" display="inline"> (N f = 104), requiring only 53,000 fitness evaluations on average. This corresponds to roughly a 50% reduction in computational complexity compared to standard genetic algorithms and enables practical real-time S-box generation with execution times as low as 8–9 seconds on modern multi-core processors.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.


