DocumentCode :
2811913
Title :
Statistical analysis of Capon-Bartlett 2-D cross spectrum
Author :
Richmond, Christ D.
Author_Institution :
Lincoln Lab., Massachusetts Inst. of Technol., Lexington, MA, USA
fYear :
2010
fDate :
14-19 March 2010
Firstpage :
2634
Lastpage :
2637
Abstract :
An exact joint probability density function (PDF) (not approximate as in) is provided for the Capon power spectral estimate and the average output power of any other deterministic filter when based on the same data sample covariance matrix. The cross coherence/cosine (correlation coefficient) between the two filter weights determines the extent of statistical dependence. An exact PDF for a sample covariance based (SCB) estimate of this cross coherence is derived allowing complete point-level statistical characterization of the 2-D Capon-Bartlett cross spectrum introduced in. The exact bias and variance are computed for the cross spectrum showing asymptotic consistency and inviting a natural recursive refinement for reducing bias.
Keywords :
correlation methods; covariance matrices; filtering theory; probability; spectral analysis; statistical analysis; Capon power spectral estimate; Capon-Bartlett 2D cross spectrum; asymptotic consistency; complete point-level statistical characterization; cosine correlation coefficient; cross coherence; deterministic filter; exact joint probability density function; filter weights; natural recursive refinement; sample covariance matrix; statistical analysis; statistical dependence; Adaptive filters; Array signal processing; Covariance matrix; Direction of arrival estimation; Filter bank; Multiple signal classification; Power generation; Probability density function; Signal to noise ratio; Statistical analysis; Adaptive; Bartlett; Capon; MVDR; SINR loss; beamforming; conventional; cross-spectra; pdf; resolution; sample coherence; sample correlation coefficient; two-dimensional;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Acoustics Speech and Signal Processing (ICASSP), 2010 IEEE International Conference on
Conference_Location :
Dallas, TX
ISSN :
1520-6149
Print_ISBN :
978-1-4244-4295-9
Electronic_ISBN :
1520-6149
Type :
conf
DOI :
10.1109/ICASSP.2010.5496258
Filename :
5496258
Link To Document :
بازگشت