基于四阶混合累积量的自适应线谱增强器

将四阶累积量的递推算法推广到求正弦信号的混合四阶累积量，并提出了一种计算正弦信号四阶混合累积量的自适应线谱增强递推算法。该线谱增强器的结构为自适应FIR滤波器，当输入为含有高斯噪声的正弦信号且处于稳态时，滤波器的权系数与输入信号的四阶混合累积量的一维切片成正比，而与高斯噪声无关，从而使信号的线谱成份得到增强。仿真结果表明，该算法从高斯色噪声中提取正弦信号的能力优于二阶自适应短时相关增强器。

A Better Adaptive Line Enhancer

Existing adaptive line enhancer (ALE) is not quite efficient for processing sinusoidal signals embedded in highly colored Gaussian noise. Existing ALE algorithm utilizes second order cumulants. The better ALE algorithm we propose is based on fourth order mixed cumulants and we call it MCBALE(Mixed Cumulants Based ALE). Section 1 derives recursive formulas which are labeled eqs.(4) through (9). The general formula, eq.(4), when applied to MCBALE, is simplified and becomes eq.(10) in section 2. Careful study of Fig.1 and eq.(11) in section 2 show that the steady state weights of the adaptive FIR(Finite Impulsive Response) filter calculated by MCBALE algorithm are independent of colored Gaussian noise. This independence is the key to the success of our MCBALE. Figs.2 through 6 give simulation results which confirm preliminarily that our proposed MCBALE is indeed better than existing ALE.