DocumentCode
3287573
Title
Integer Levinson algorithms for Toeplitz and certain Toeplitz-like matrices
Author
Bistritz, Y. ; Segalov, Y.
Author_Institution
Sch. of Electr. Eng., Tel Aviv Univ., Tel Aviv, Israel
fYear
2010
fDate
June 30 2010-July 2 2010
Firstpage
5720
Lastpage
5725
Abstract
The paper presents an integer Levinson algorithm for certain Toeplitz-like (quasi-Toeplitz) matrices. The integer preserving (IP) property means that for a Toeplitz matrix with (complex or real) integers, the algorithm is completed over integers without encountering quotients. The algorithm also produces triangular factorization of the inverse matrix with integer matrices. The derivation begins with an intermediate algorithm that is IP simply because it is division-free but it produces integers whose size increases at a severe rate. Next, the main algorithm is obtained by identifying and recursively dividing out common integers that the division-free algorithm is shown to produce systematically. The result is an efficient integer algorithm with integers of least size. This way of derivation also provides a constructive proof for the IP property of the algorithm. The integer Levinson algorithms for a non-symmetric (real or complex) Toeplitz is deduced from the more general main result from where the integer algorithm for the Hermitian Toeplitz case follows readily.
Keywords
Toeplitz matrices; matrix decomposition; signal processing; Integer Levinson algorithms; Toeplitz-like matrices; division-free algorithm; integer preserving; inverse matrix; triangular factorization; Equations; Finite impulse response filter; Nonhomogeneous media; Predictive models; Signal processing algorithms; Speech processing; Stability; Symmetric matrices; Testing; USA Councils;
fLanguage
English
Publisher
ieee
Conference_Titel
American Control Conference (ACC), 2010
Conference_Location
Baltimore, MD
ISSN
0743-1619
Print_ISBN
978-1-4244-7426-4
Type
conf
DOI
10.1109/ACC.2010.5531143
Filename
5531143
Link To Document