基于多路延迟结构的频率估计新算法研究

Research on a New Algorithm for Frequency Estimation Based on Multi-path Delay Structure

针对多路延迟数字测频结构,提出了基于多级维纳滤波(MSWF)的MUSIC频率估计方法。该方法利用MSWF的前向递推的多级分解特性来实现信号子空间和噪声子空间的快速估计,不需要估计协方差矩阵和对其作特征值分解,所以运算复杂度可以明显降低,并且子空间估计性能与常规方法几乎一样。最后在子空间分解的基础上,通过MUSIC算法对其频率估计性能进行比较。仿真结果表明MSWF-MUSIC算法和常规子空间分解MUSIC算法具有几乎一样的频率估计精度;但本文算法降低了计算量和复杂度,更易于信号的实时处理;随信噪比或快拍数的增加,两算法的频率估计误差都越来越趋近于Cramer-Rao界。

To consider the digital frequency measurement structure of multi-path delay, a MUSIC frequency estimation algorithm based on the multi-stage Wiener filter （MSWF） was presented. The method can get signal subspace and noise subspace quickly by using multi-decomposition of the forward recursions of the MSWF, which does not require the formation of the covariance matrix and its eigenvalue decomposition, to reduce the computational complexity significantly. The novel approach can achieve such subspace estimation performance as that of the classical method. On the basis of subspace decomposition, the frequency estimation performance of MSWF-MUSIC algorithm was compared with that of MUSIC one of the classical method by the simulation. The simulated results show that both algorithms have almost the same frequency estimation accuracy; the computational amount and its complexity are decreased significantly to process the signal real time and easily; with the increment of SNR of snapshot, the frequency estimation mean square errors of both algorithms approach to Cramer-Rao boundary more and more.