DocumentCode
2905521
Title
Exact Expressions for the Condition Number Distribution of Complex Wishart Matrices
Author
Matthaiou, Michail ; McKay, Matthew R. ; Smith, Peter J. ; Nossek, Josef A.
Author_Institution
Inst. for Circuit Theor. & Signal Process., Tech. Univ. Munchen (TUM), Munich, Germany
fYear
2010
fDate
23-27 May 2010
Firstpage
1
Lastpage
6
Abstract
The standard condition number (SCN) is a fundamental metric in the context of multiple-input multiple-output communication systems, linear detection and classical linear algebra. Hence, in this paper we propose a novel generic framework for the SCN distribution of three different classes of Wishart matrices which leads to new results and insights. In particular, our analysis covers both central and non-central Wishart distributions of arbitrary dimension and therefore is applicable to uncorrelated/semi-correlated Rayleigh fading and Ricean fading scenarios. For the special case of dual semi-correlated central Wishart matrices, we derive exact and asymptotic polynomial expressions for the SCN distributions. All analytical results are validated via Monte-Carlo simulations with the attained accuracy being excellent in all cases. The impact of the model parameters on channel conditioning is also investigated in detail.
Keywords
MIMO communication; Monte Carlo methods; Rayleigh channels; matrix algebra; number theory; Monte-Carlo simulations; Ricean fading; asymptotic polynomial expressions; classical linear algebra; complex Wishart matrices; condition number distribution; linear detection; multiple-input multiple-output communication systems; semi-correlated Rayleigh fading; standard condition number; uncorrelated Rayleigh fading; Circuit theory; Communications Society; Context; Covariance matrix; Eigenvalues and eigenfunctions; MIMO; Polynomials; Probability density function; Rayleigh channels; Signal Processing Society;
fLanguage
English
Publisher
ieee
Conference_Titel
Communications (ICC), 2010 IEEE International Conference on
Conference_Location
Cape Town
ISSN
1550-3607
Print_ISBN
978-1-4244-6402-9
Type
conf
DOI
10.1109/ICC.2010.5502235
Filename
5502235
Link To Document