The Covariance Least-Squares Algorithm for Spectral Estimation of Processes of Short Data Length

摘要: A new method for generating the autoregressive (AR) process parameters spectral estimation is introduced. The fits AR models to data optimally in sense of minimizing sum squares error covariance function within model prediction region, and thus designated as Covariance Least-Squares (CLS) algorithm. This minimization shown be identical with weighted average one-step, linear errors adaptive weights corresponding energy region. CLS algorithm compared (LS) [1], [2] by simulation asymptotic properties. It that combines all desirable properties comparison improved robustness presence nonstationarity, namely, additive transients envelope modulation. also provides asymptotically unbiased parameters, a property shared LS