前馈网络的频谱分析与算法

Spectral Analysis and Algorithm on Nonlinear Feedforward Generators

本文讨论对于前馈网络,如何选取抽头序列的相位向量集,使得前馈序列在Galois域Fourier变换下某一频率分量为零(即前馈序列的线性复杂度退化)。当LFSR的级数n和前馈函数的变元个数m互素时,本文由PFZ算法给出了上述问题的存在性、计数和构造性组合问题的完整解,且证明了当n或m较小时,PFZ算法是一个有效的算法。

The problem,how to choose the set of phase vectors of tap sequences in the nonlinear feedforward generator such that some frequency component of the feedforward sequence under Galois-Field Fourier transform is equal to zero, will be discussed in this article. We will give the solution of the problem's existence, enumeration and construction by PFZ Algorithm when the number n of the LFSR's stages and the number m of variables of the feedforward function are coprime, and prove PFZ Algorithm is efficient when n or m is small.