Skip to Main Content
In the letter the performance of the Fougere method for autoregressive (AR) spectral analysis is studied. This method is compared with the Burg method and the least-squared (LS) method (Marple algorithm). It is shown that when the signal consists of two equal-level sinusoids in white noise and the data are complex-valued, as they will be in antenna applications, the Fougere method improves the results obtained by the Burg method in all cases. Also, for low signal/noise ratio (SNR), the Fougere method gives better results than LSM.