Neural Network Chaotic Encryption Algorithm And Fpga Design

With the development of computer and internet technology, more and more information needs to be transported via the open channel which is insecurity. So, how to keep the information secret has already become a hot focus. Encryption is an important way to achieve information security. However, most typical encryption algorithms can not resist the attacks from cipher experts or interceptors. It must increase the number of secret-key to guarantee its security, but, the cost of this is the increase of computation complexity and the decrease of encryption speed. Therefore, looking for a quick, safe and efficient encryption algorithm becomes a trend in the current cryptography. However, chaotic neural network owns many properties which have close relations with cryptography. Especially, due to its massively parallel computing nature and chaotic property, the chaotic neural network ideally suits for encryption, and the method based on that becomes a new design thought in cryptography.Necessary knowledge is introduced at the beginning of this thesis, and analysis about the security of various algorithms is also made. Then, the nonlinear dynamics and parallel processing nature of Hopfield Neural Network (HNN) are analyzed to prove that it will have good prospects in information security technology, because of its property of fast parallel real-time processing and high complex computation. We have deeply researched two encryption algorithms based on the chaotic nature of HNN, and improved them, they are: 1) Symmetric stream cryptosystem based on chaotic attractors, in that algorithm, the nonlinear outputs of some linear feedback registers are selected as the output of the whole scheme by discrete HNN, and that process forms a kind of stream algorithms with considerable complexity. 2) Public-key cryptosystem based on chaotic attractors, a kind of the Diffie-Hellman public-key cryptosystem associated with symmetric encryption. The random synaptic matrix is kept as secret key and the permuted matrix as public key to implement key exchange. On that foundation, we propose the Hybrid cryptosystem based on chaotic attractors, which used public-key cryptosystem based on chaotic attractors to manage key and generate random sequence, while symmetric-key algorithm used the random sequence to encrypt the plaintext. It not only has the easy distribution of secret-key characteristics in public key cryptosystem, but also has the high speed computation nature of symmetric cryptography, and needn't enter the symmetric encryption key from the outside world.At the end, an IP core is designed to implement the public-key cryptosystem based on chaotic attractor, which has been improved in this paper. In this work, these following things are mainly done on FPGA: dividing the whole scheme, describing each module on RTL level in Hardware Description Language, and compiling and simulating the whole scheme. The simulation result shows that this algorithm has an acceptable encryption speed.