摘要
针对二维衰减谐波估计方法在大样本容量条件下计算量过大的问题,提出了一种基于二维四阶混合累积量的ESPRIT方法。在不影响估计精度的前提下,通过保持特征矩阵结构不变,有效地减小了算法的计算量。利用四阶混合累积量对高斯噪声的自动抑制,降低了参数估计的信噪比门限并提高了估计精度。仿真实验表明,在高斯色噪声条件下该方法的计算效率和估计性能优于现有方法。
A new method for two-dimensional damped sinusoids estimation in colored Gaussian noise, called two-dimensional fourth-order mixed cumulants based on ESPRIT (2d-FOMCESPRIT), is presented. The amount of computations is greatly reduced by holding the dimensions of the feature matrix compared with that of the former methods in the presence of a long record. Due to FOMC's auto suppressing Gaussian noise, the SNR threshold is reduced and the precision of the parameter estimation is improved. Simulation results show that this method is of better performance and higher efficiency than the former methods in colored Gaussian noise.
出处
《国防科技大学学报》
EI
CAS
CSCD
北大核心
2005年第1期55-59,共5页
Journal of National University of Defense Technology
基金
国家自然科学基金资助项目(69901005)
关键词
四阶混合累积量
二维衰减谐波
谐波恢复
高斯色噪声
Fourth-Order Mixed Cumulants(FOMC)
two-dimensional damped sinusoids
harmonic retrieval
colored Gaussian noise
