Capon and APES Spectrum Estimation for Real-valued Signals

Andreas Jakobsson , Torbjörn Ekman and Petre Stoica

Eighth IEEE Digital Signal Processing Workshop,
Bryce Canyon National Park, Utah, Aug. 8-12 1998.   © 1998 IEEE

Abstract:

This paper considers the problem of estimating the spectrum of real-valued signals. We propose real-valued versions of the Capon and the APES spectrum estimators. The estimators are derived as members of the Matched-Filterbank (MAFI) estimator class introduced in [Stoica et. al., Signal Processing, 66(1)].

Furthermore, we show that the real-valued estimators will be unbiased, whereas the complex-valued estimators will have a (slight) bias for real-valued data. Finally, we conclude the paper with a numerical example illustrating the performance of the proposed estimator.

Related publications:

2-D Capon Method, ICASSP 2000-paper.

1-D Capon Method, EUSIPCO 2000-paper.

Source:

Pdf, 104K