一种抑制符号速率估计背景色噪声的非线性滤波算法

数字调制信号符号速率估计的依据是循环平稳理论,由于信号的符号速率就是其基本循环频率,因此可以通过提取信号非线性变换(例如循环自相关函数)的循环频率获知符号速率.但是,非线性变换不仅能产生对应于符号速率的正弦分量及其各次谐波,还会将信号自身转变成不利于谱线提取的连续有色噪声(其能量主要分布在低频部分).当观察数据长度有限时,自噪声对谱线提取的影响尤其明显.本文深入研究了数字调制信号非线性变换的频域特征,充分利用离散频率分量不同于连续噪声而在其邻域内突起的显著特点,提出一种能够有效抑制背景色噪声的非线性滤波算法.文中详尽的Monte Carlo仿真验证了算法的有效性.

A New Nonlinear Filtering Algorithm for Colored Background Self-Noise Suppressing of Symbol Rate Estimation

According to the theory of cyclostationary,symbol rate of some digital modulated signals,whose symbol rate gives out basic cyclic frequency,can be retrieved by estimating cyclic frequencies of signal's nonlinear transform such as cyclic autocorrelation.Nonlinear transform of signal will generate symbol-rate sinusoidal wave and its harmonics as well as continuous colored background self-noise,which distribute mainly within low frequency band and can disturb the identification of spectral lines especially when available data is not enough.This paper delves into the frequency characteristic of nonlinear transform and proposes a novel nonlinear filtering algorithm,which can be utilized to suppress continuous colored background self-noise,based on an obvious fact that symbolrate spectral line will stand out within its nearby frequency domain,a key advantage that will never belong to continuous self-noise.All-sided Monte Carlo simulations have been carried out to justify this algorithm.