基于ESPRIT的噪声抑制频率估计算法

在中低信噪比时，协方差矩阵受噪声影响较大导致ESPRIT算法性能降低，使其与克拉美罗下限(CRLB)有一定距离。针对该问题，提出一种基于ESPRIT的噪声抑制频率估计算法，利用信号频域内若干子带的谱线估计协方差矩阵，通过该矩阵的特征向量张成信号子空间，估计信号各分量的频率。实验结果表明，该算法能用于多个频率分量的信号分析，归一化频率估计的范围为 ，且性能接近于CRLB下限。

Noise Suppression Frequency Estimation Algorithm Based on ESPRIT

In low to medium Signal to Noise Ratio(SNR), ESPRIT algorithm has low performance for noise affecting covariance matrix, and it makes ESPRIT algorithm have a performance gap to Cremer-Rao Lower Bound(CRLB). Aiming at this problem, this paper presents a noise suppression frequency estimation algorithm based on ESPRIT. It estimates the covariance matrix with spectral lines inside several sub-bands spectrum, signal subspace is spanned by the generalized eigenvectors of the improved covariance matrix and frequency is simultaneously resolved with the signal subspace. Experimental results show that this algorithm can be applied in multiple-frequency signals, it has full frequency range of , and the performance approaches CRLB.