摘要
Power spectrum estimation is to use the limited length of data to estimate the power spectrum of the signal. In this paper, we study the recently proposed tunable high-resolution estimator（THREE）, which is based on the best approximation to a given spectrum, with respect to different notions of distance between power spectral densities. We propose and demonstrate a different distance for the optimization part to estimate the multivariate spectrum. Its effectiveness is tested through Matlab simulation. Simulation shows that our approach constitutes a valid estimation procedure. And we also demonstrate the superiority of the method, which is more reliable and effective compared with the standard multivariate identification techniques.
Power spectrum estimation is to use the limited length of data to estimate the power spectrum of the signal. In this paper, we study the recently proposed tunable high-resolution estimator（THREE）, which is based on the best approximation to a given spectrum, with respect to different notions of distance between power spectral densities. We propose and demonstrate a different distance for the optimization part to estimate the multivariate spectrum. Its effectiveness is tested through Matlab simulation. Simulation shows that our approach constitutes a valid estimation procedure. And we also demonstrate the superiority of the method, which is more reliable and effective compared with the standard multivariate identification techniques.
基金
supported by the National Natural Science Foundation of China (61379014)
