2009 IEEE 10th Workshop on Signal Processing Advances in Wireless Communications 2009
DOI: 10.1109/spawc.2009.5161847
|View full text |Cite
|
Sign up to set email alerts
|

Adjusting the generalized likelihood ratio test of circularity robust to non-normality

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
2
1
1
1

Citation Types

0
18
0

Year Published

2009
2009
2024
2024

Publication Types

Select...
5
3

Relationship

0
8

Authors

Journals

citations
Cited by 25 publications
(18 citation statements)
references
References 14 publications
0
18
0
Order By: Relevance
“…With the previous notations, the sphericity test makes use of the GLRT adjusted for CES (Complex Elliptically Symmetric) distributions [7], [8], known as aGLRT.…”
Section: Spherically Invariant Random Processesmentioning
confidence: 99%
“…With the previous notations, the sphericity test makes use of the GLRT adjusted for CES (Complex Elliptically Symmetric) distributions [7], [8], known as aGLRT.…”
Section: Spherically Invariant Random Processesmentioning
confidence: 99%
“…The moments (19), which fully determine the null distribution of T , and hence of M , do not depend on the true covariance matrix ⌃ r . Hence we can assume ⌃ r = I 4p when simulating M .…”
Section: G Direct Simulationmentioning
confidence: 99%
“…As with the LRT for complex propriety [19], our LRT for quaternion propriety will be sensitive to departures from Gaussianity which could be an issue when applying the test to physical data.…”
Section: Asymptotic Robustness To Nongaussianitymentioning
confidence: 99%
“…A general test of circularity, for the class of elliptically symmetric distributions, is presented in [13] assuming a known covariance matrix. A complex GLRT for this distribution is presented in [14] which adjusts the test statistic by an estimate of the fourth moment to handle the non-Gaussian case. This adjustment, however, may be sensitive to outliers and requires many samples to estimate.…”
Section: Introductionmentioning
confidence: 99%
“…The CGGD adapts to a large family of symmetric distributions, from super-Gaussian to sub-Gaussian including specific densities such as bivariate Laplacian and Gaussian distributions. Since the CGGD is also a member of the elliptically symmetric distributions, the normalized kurtosis values of the real and imaginary parts of a complex random variable are a scaled version of the complex kurtosis where the scale factor is nonnegative and is a function of noncircularity as shown in [14]. Since the kurtosis of the complex Gaussian is zero, as in the realvalued case, positive normalized kurtosis values imply a superGaussian distribution, i.e., a sharper peak with heavier tails, and negative normalized kurtosis values imply sub-Gaussian distributions.…”
Section: Introductionmentioning
confidence: 99%