Inversion of Parahermitian matrices

Stephan Weiss, Andrew P. Millar, Robert W. Stewart

Research output: Contribution to conferencePaper

6 Citations (Scopus)
61 Downloads (Pure)

Abstract

Parahermitian matrices arise in broadband multiple-input multiple-output (MIMO) systems or array processing, and require inversion in some instances. In this paper, we apply a polynomial eigenvalue decomposition obtained by the sequential best rotation algorithm to decompose a parahermitian matrix into a product of two paraunitary, i.e.lossless and easily invertible matrices, and a diagonal polynomial matrix. The inversion of the overall parahermitian matrix therefore reduces to the inversion of auto-correlation sequences in this diagonal matrix. We investigate a number of different approaches to obtain this inversion, and and assessment of the numerical stability and complexity of the inversion process.
Original languageEnglish
Pages447-451
Number of pages5
Publication statusPublished - 23 Aug 2010
Event18th European Signal Processing Conference - Aalborg, Denmark
Duration: 23 Aug 201028 Aug 2010

Conference

Conference18th European Signal Processing Conference
CityAalborg, Denmark
Period23/08/1028/08/10

Fingerprint

matrix
eigenvalue
autocorrelation
inversion
decomposition

Keywords

  • parahermitian matrices
  • inversion
  • electrical engineering
  • signal processing

Cite this

Weiss, S., Millar, A. P., & Stewart, R. W. (2010). Inversion of Parahermitian matrices. 447-451. Paper presented at 18th European Signal Processing Conference, Aalborg, Denmark, .
Weiss, Stephan ; Millar, Andrew P. ; Stewart, Robert W. / Inversion of Parahermitian matrices. Paper presented at 18th European Signal Processing Conference, Aalborg, Denmark, .5 p.
@conference{a989bcd23a6f45e6bcbce0bea92672bc,
title = "Inversion of Parahermitian matrices",
abstract = "Parahermitian matrices arise in broadband multiple-input multiple-output (MIMO) systems or array processing, and require inversion in some instances. In this paper, we apply a polynomial eigenvalue decomposition obtained by the sequential best rotation algorithm to decompose a parahermitian matrix into a product of two paraunitary, i.e.lossless and easily invertible matrices, and a diagonal polynomial matrix. The inversion of the overall parahermitian matrix therefore reduces to the inversion of auto-correlation sequences in this diagonal matrix. We investigate a number of different approaches to obtain this inversion, and and assessment of the numerical stability and complexity of the inversion process.",
keywords = "parahermitian matrices, inversion, electrical engineering, signal processing",
author = "Stephan Weiss and Millar, {Andrew P.} and Stewart, {Robert W.}",
year = "2010",
month = "8",
day = "23",
language = "English",
pages = "447--451",
note = "18th European Signal Processing Conference ; Conference date: 23-08-2010 Through 28-08-2010",

}

Weiss, S, Millar, AP & Stewart, RW 2010, 'Inversion of Parahermitian matrices' Paper presented at 18th European Signal Processing Conference, Aalborg, Denmark, 23/08/10 - 28/08/10, pp. 447-451.

Inversion of Parahermitian matrices. / Weiss, Stephan; Millar, Andrew P.; Stewart, Robert W.

2010. 447-451 Paper presented at 18th European Signal Processing Conference, Aalborg, Denmark, .

Research output: Contribution to conferencePaper

TY - CONF

T1 - Inversion of Parahermitian matrices

AU - Weiss, Stephan

AU - Millar, Andrew P.

AU - Stewart, Robert W.

PY - 2010/8/23

Y1 - 2010/8/23

N2 - Parahermitian matrices arise in broadband multiple-input multiple-output (MIMO) systems or array processing, and require inversion in some instances. In this paper, we apply a polynomial eigenvalue decomposition obtained by the sequential best rotation algorithm to decompose a parahermitian matrix into a product of two paraunitary, i.e.lossless and easily invertible matrices, and a diagonal polynomial matrix. The inversion of the overall parahermitian matrix therefore reduces to the inversion of auto-correlation sequences in this diagonal matrix. We investigate a number of different approaches to obtain this inversion, and and assessment of the numerical stability and complexity of the inversion process.

AB - Parahermitian matrices arise in broadband multiple-input multiple-output (MIMO) systems or array processing, and require inversion in some instances. In this paper, we apply a polynomial eigenvalue decomposition obtained by the sequential best rotation algorithm to decompose a parahermitian matrix into a product of two paraunitary, i.e.lossless and easily invertible matrices, and a diagonal polynomial matrix. The inversion of the overall parahermitian matrix therefore reduces to the inversion of auto-correlation sequences in this diagonal matrix. We investigate a number of different approaches to obtain this inversion, and and assessment of the numerical stability and complexity of the inversion process.

KW - parahermitian matrices

KW - inversion

KW - electrical engineering

KW - signal processing

UR - http://www.eurasip.org/Proceedings/Eusipco/Eusipco2010/Contents/papers/1569295849.pdf

M3 - Paper

SP - 447

EP - 451

ER -

Weiss S, Millar AP, Stewart RW. Inversion of Parahermitian matrices. 2010. Paper presented at 18th European Signal Processing Conference, Aalborg, Denmark, .