Design Of S-box Based On Chaos And Computational Intelligence Algorithm

Block cipher which is one important branch of modern cryptography has the advantages of fast running speed, easy to be standardized and easy realization of software and hardware. It has been widely used in computer communication and information security. It has become one of the hot and difficult topic in cryptography. S-box is the only nonlinear component in lots of block cipher algorithms. It plays an important role in confusion. It makes the relationship between the key and the plaintext, the key and the ciphertext complicated. How to design a safe and effective S-box is the key and difficult problems in the design of block cipher algorithm.Chaos has the characteristics of ergodicity, nonlinear, initial value sensitivity and randomness. The characteristic of chaos and cryptography in the encryption system was similarity, and there are also some contact. We can construct good S-boxes with good cryptographic properties in chaotic system.Computational intelligent algorithm includes a lot of knowledge in many disciplines of mathematics, the theory of biological evolution, artificial intelligence and physics. It provides a new idea of how to solve many complex problems. Using the method of computational intelligence algorithm for constructing S-boxes also made some research achievements at present.The methods of constructing S-boxes based on single chaotic map have the disadvantages of great randomness, more difficult to search to the S-boxes with good cryptographic properties. Aiming at these shortcomings, two candidate design methods of S-boxes has been proposed based on chaotic neural network and chaotic genetic algorithm. The performance analysis and comparison of S-boxes proved that: The methods of constructing S-boxes based on the combination of chaos and neural network algorithm, chaos and genetic algorithm have good cryptology characteristic. The main research contents are as follows:First,this paper studies the structure, model, characteristics and learning method of the neural network. Neural network has the characteristics of parallel search, nonlinear mapping and associative memory. It has a great advantage in searching the global optimal solution. It can overcome the disadvantages of the algorithm of chaos when neural network and chaos algorithm is combined with the structure of S-boxes. Because of these reasons, a method is proposed based on chaos algorithm and neural network to construct S-boxes. And, this paper analyses and compares the performance of the cryptography through the experiment.Second, this paper studies the concept, principle and algorithm process of genetic algorithm. Chaotic variables have the characteristics of randomness, ergodicity and regularity. It can improve the efficiency of lobal searching and overcome the defects of genetic algorithm when chaos algorithm and genetic algorithm are combined to construct S-boxes. Because of these reasons, an approach is proposed based on chaos algorithm and genetic algorithm to construct S-boxes. And, this paper analyses and compares the performance of the cryptography through the experiment.Third, this paper studies the composition of communication network for CAPF and the structure model of security transmission system of communication network for CAPF. Hybrid encryption algorithm AES-RSA based on the candidate S-boxes has been proposed base on the approach of construct S-boxes by chaotic neural network and genetic algorihm of chaos. And, this paper verify the encryption performance by encryption experiment. Finally, the performance of hybrid encryption has been analysed by the security transmission system of communication network for CAPF.