基于多层前馈神经网络的分组密码设计

DESIGN OF BLOCK CIPHER BASED ON MULTY-LAYER FEED-FORWARD NEURAL NETWORK

基于多层前馈神经网络的特性和分组密码的设计原则,构造了一种分组密码的数学模型,并用一个两层前馈网络具体实现了该分组密码体制。通过仿真,说明了该分组密码体制是可行的;通过对其安全性进行分析并与DES相比较,说明该分组密码体制具有较高的安全性,具有很好的混乱特征和扩散特征,并易于实现。

This article introduces the construction of the mathematical model of a block cipher based on the feature of multi-layer feed-forward NN and on the block cipher design principle, and a particular implementation of the block cipher system on a two-layer feed-forward network. The simulation indicated that the system is feasible. By the analyses of its security and in comparison with DES the block cipher system presented a high degree of security and was good at its confusion and diffusion features as well as to be easily implemented.