Abstract-We provide a frame-theoretic analysis of oversampled finite impulse response (FIR) and infinite impulse response (IIR) uniform filter banks (FB's). Our analysis is based on a new relationship between the FB's polyphase matrices and the frame operator corresponding to an FB. For a given oversampled analysis FB, we present a parameterization of all synthesis FB's providing perfect reconstruction. We find necessary and sufficient conditions for an oversampled FB to provide a frame expansion. A new frame-theoretic procedure for the design of paraunitary FB's from given nonparaunitary FB's is formulated. We show that the frame bounds of an FB can be obtained by an eigenanalysis of the polyphase matrices. The relevance of the frame bounds as a characterization of important numerical properties of an FB is assessed by means of a stochastic sensitivity analysis. We consider special cases in which the calculation of the frame bounds and synthesis filters is simplified. Finally, simulation results are presented.
In the study of Weyl-Heisenberg frames the assumption of having a finite frame upper bound appears recurrently. In this article it is shown that it actually depends critically on the time-frequency lattice used. Indeed, for any irrational α > 0 we can construct a smooth g ∈ L 2 (R) such that for any two rationals a > 0 and b > 0 the collection (g na,mb ) n,m∈Z of time-frequency translates of g has a finite frame upper bound, while for any β > 0 and any rational c > 0 the collection (g ncα,mβ ) n,m∈Z has no such bound. It follows from a theorem of I. Daubechies, as well as from the general atomic theory developed by Feichtinger and Gröchenig, that for any nonzero g ∈ L 2 (R) which is sufficiently well behaved, there exist a c > 0, b c > 0 such that (g na,mb ) n,m∈Z is a frame whenever 0 < a < a c , 0 < b < b c . We present two examples of a nonzero g ∈ L 2 (R), bounded and supported by (0, 1), for which such numbers a c , b c do not exist. In the first one of these examples, the frame bound equals 0 for all a > 0, b > 0, b < 1. In the second example, the frame lower bound equals 0 for all a of the form l · 3 −k with l, k ∈ N and all b, 0 < b < 1, while the frame lower bound is at least 1 for all a of the form (2m) −1 with m ∈ N and all b, 0 < b < 1.
A B S T R A C TWe provide a frame-theoretic analysis of oversampled and critically sampled, FIR and IIR, uniform filter banks (FBs). Our analysis is based on a relation between the polyphase matrices and the frame operator. For a given oversampled analysis FB, we parameterize all synthesis FBs providing perfect reconstruction, and we discuss the minimum norm synthesis FB and its approximative construction. We find conditions for a FB to provide a frame expansion. Paraunitary and biorthogonal FBs are shown to correspond to tight and exact frames, respectively. A new procedure for the design of paraunitary FBs is formulated. We show that the frame bounds are related with the eigenvalues of the polyphase matrices and the oversampling factor, and that they determine important numerical properties of the FB. This paper presents a new frame-theoretic approach to oversampled and critically sampled FIR and IIR FBs. Our approach, which extends an analysis of continuous-time Weyl-Heisenberg frames proposed in [13], is based on an important relation between the FB's polyphase matrices and the frame operator. In Section 2, we review FBs, introduce the corresponding type of frames, and show that the FB's polyphase matrices provide matrix representations of the frame operator. Section 3 shows that the frame bounds determine important numerical properties of FBs, and that they are related to the eigenvalues of the polyphase matrices and the oversampling factor. In Section 4, we parameterize all synthesis FBs providing perfect reconstruction (PR) for a given oversampled analysis FB, and we discuss the minimum norm synthesis FB and its approximative construction. Conditions for a FB to provide a frame expansion are formulated, and the equivalence of critically sampled (biorthogonal) FBs and exact frames is discussed. In Section 5, we show that paraunitary FBs correspond to tight frames, and we propose a method for constructing paraunitary FBs from given nonparaunitary FBs. 'For the sake of brevity, we shall henceforth use the term filter bank (FB) instead of uniform filter bank. I N T R O D U C T I O N2We note that our theory can easily be extended to PR with nonzero delay.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
customersupport@researchsolutions.com
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
This site is protected by reCAPTCHA and the Google Privacy Policy and Terms of Service apply.
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.