自回归滑动平均(ARMA)谱估计的广义最小二乘法(GLS)算法及其在雷达频率起伏特性测量中的应用

A GLS Algorithm for ARMA Spectral Estimation and Its Application to Radar Signal Processing

提出一种自回归滑动平均(ARMA)谱估计的广义最小二乘法算法,它利用白化滤波器将有色噪声转化成白噪声,并通过迭代方式改进参数估计。还提出超松弛法和综合判阶技术。本算法主要适用于正弦信号加白噪声(SWN),以及受白噪声污染的自回归过程(NAR)。统计分析和实验表明,本算法具有分辨率高、频率偏移小等优点。

A generalized least squares (GLS) method for ARMA spectral estimation is presented. The procedure uses a whitening filter to convert the colored noise to a white noise, and an iterative method to improve the parameter estimation. An overrelaxation technique and a hybrid order determination technique are proposed. The GLS algorithm presented in this paper is suitable mainly to the process of sinusoids plus white noise (SWN) and the noisy autoregressive(NAR) process. The stati- stic analysis and experiments show that the proposed algorithm has the advantage of sharper resolution and less frequency bias.