Abstract
This chapter constitutes the second part in a series of papers on distributed noise-shaping quantization. In the first part, the main concept of distributed noise shaping was introduced and the performance of distributed beta encoding coupled with reconstruction via beta duals was analyzed for random frames (Chou and Güntürk, Constr Approx 44(1):1–22, 2016). In this second part, the performance of the same method is analyzed for several classical examples of deterministic frames. Particular consideration is given to Fourier frames and frames used in analog-to-digital conversion. It is shown in all these examples that entropic rate-distortion performance is achievable.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
References
R. Adler, T. Nowicki, G. Świrszcz, C. Tresser, Convex dynamics with constant input. Ergodic Theory Dyn. Syst. 30, 957–972 (2010)
J.J. Benedetto, A.M. Powell, Ö. Yılmaz, Sigma-delta quantization and finite frames. IEEE Trans. Inf. Theory 52(5), 1990–2005 (2006)
J.J. Benedetto, O. Oktay, A. Tangboondouangjit, Complex sigma-delta quantization algorithms for finite frames, in Radon Transforms, Geometry, and Wavelets. Contemporary Mathematics, vol. 464 (American Mathematical Society, Providence, RI, 2008), pp. 27–49
B.G. Bodmann, V.I. Paulsen, Frame paths and error bounds for sigma-delta quantization. Appl. Comput. Harmon. Anal. 22(2), 176–197 (2007)
P.G. Casazza, S.J. Dilworth, E. Odell, Th. Schlumprecht, A. Zsk, Coefficient quantization for frames in Banach spaces. J. Math. Anal. Appl. 348(1), 66–86 (2008)
E. Chou, C.S. Güntürk, Distributed noise-shaping quantization: I. Beta duals of finite frames and near-optimal quantization of random measurements. Constr. Approx. 44(1), 1–22 (2016)
I. Daubechies, R. DeVore, Approximating a bandlimited function using very coarsely quantized data: a family of stable sigma-delta modulators of arbitrary order. Ann. Math. 158(2), 679–710 (2003)
C.S. Güntürk, Mathematics of analog-to-digital conversion. Comm. Pure Appl. Math. 65(12), 1671–1696 (2012)
A.M. Powell, R. Saab, Ö. Yılmaz, Quantization and finite frames, in Finite Frames: Theory and Applications, ed. by G.P. Casazza, G. Kutyniok. Applied and Numerical Harmonic Analysis (Birkhäuser/Springer, New York, 2013), pp. 267–302
R. Schreier, G.C. Temes, Understanding Delta-Sigma Data Converters (Wiley/IEEE Press, Chichester, 2004)
Y. Wang, Sigma-delta quantization errors and the traveling salesman problem. Adv. Comput. Math. 28(2), 101–118 (2008)
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Appendix: Greedy Quantizer for Complex Measurements
Appendix: Greedy Quantizer for Complex Measurements
In this section we will provide a generalization of the complex-valued ΣΔ quantization algorithm given in [3, Proposition 3.1] and the greedy noise-shaping quantization algorithm given in [6, Theorem 2.1]. The result, which is applicable to both real and complex quantization alphabets, offers nontrivial improvements in the complex case, thanks to the use of general semi-norms to measure closeness.
Lemma 2
Let \(\mathscr{A}\) be a quantization alphabet in \(\mathbb{C}\) , B ∗ be the closed unit ball of a semi-norm | ⋅ |∗ on \(\mathbb{C}\) treated as a vector space over \(\mathbb{R}\) , and H: = (H n, m ) n, m ∈ [N] be an N × N real-valued lower-triangular matrix with unit diagonal. Suppose there exist positive real numbers μ, δ, γ such that
and
Then for any \(y \in \mathbb{C}^{N}\) such that | y n |∗ ≤ μ for all n ∈ [N], there exist \(q \in \mathscr{ A}^{N}\) and \(u \in \mathbb{C}^{N}\) such that
where | u n |∗ ≤ δ for all n ∈ [N].
Proof
The proof of this result is yet another adaptation of a well-known induction argument. By our assumption on H, we are seeking to satisfy the equations
for all n ∈ [N].
Since | y 0 |∗ ≤ μ ≤ γ, (27) implies that there exist \(q_{0} \in \mathscr{ A}\) and u 0 ∈ δB ∗ such that u 0 + q 0 = y 0. Hence (29) is satisfied for n = 0 and | u 0 |∗ ≤ δ.
For the induction step, assume that | u m |∗ ≤ δ for all m < n, and let
Using sub-additivity and homogeneity of | ⋅ |∗ followed by the condition given in (28), we get
hence, because of (27) again, there exist \(q_{n} \in \mathscr{ A}\) and u n ∈ δB ∗ such that u n + q n = w n , i.e. (29) holds. □
Special known cases. There are certainly many ways to choose \(\mathscr{A}\) and | ⋅ |∗. We first note two important special cases of practical importance. Here L denotes \(\vert \mathscr{A}\vert\).
-
\((\mathbb{R})\) Real arithmetic progression
This quantizer uses \(\mathscr{A}:=\mathscr{ A}_{L,\delta }:=\{ (-L + 2l - 1)\delta: 1 \leq l \leq L\} \subset \mathbb{R}\), i.e. the origin-symmetric arithmetic progression of length L and spacing 2δ along with \(\vert z\vert _{{\ast}}:= \vert \mathfrak{R}(z)\vert\). Then B ∗ is the infinite vertical strip \(\{z: \vert \mathfrak{R}(z)\vert \leq 1\}\) and (27) holds for γ: = Lδ. Using the algorithm in Lemma 2, \(y \in \mathbb{R}^{N}\) results in \(u \in \mathbb{R}^{N}\), and ∥y∥ ∞ ≤ μ implies ∥u∥ ∞ ≤ δ so that the setup becomes identical to that of [6].
-
\((\mathbb{C})\) Complex square lattice quantizer
This quantizer assumes L = K 2 for some positive integer K and sets \(\mathscr{A}:=\mathscr{ A}_{K,\delta } + i\mathscr{A}_{K,\delta } \subset \mathbb{C}\) along with \(\vert z\vert _{{\ast}}:=\max (\vert \mathfrak{R}(z)\vert,\vert \mathfrak{I}(z)\vert )\). B ∗ can be identified with [−1, 1]2 (as a subset \(\mathbb{R}^{2}\)) so that (27) is valid for γ: = Kδ. Since \(\vert z\vert _{{\ast}}\leq \vert z\vert \leq \sqrt{2}\vert z\vert _{{\ast}}\) for any \(z \in \mathbb{C}\), ∥y∥ ∞ ≤ μ implies | y n |∗ ≤ μ for all n and Lemma 2 then yields \(\|u\|_{\infty }\leq \sqrt{2}\delta\).
When K is even, the resulting \(\mathscr{A}\) has no real points and it may be desirable to require that \(y \in \mathbb{R}^{N}\) always yields \(q \in \mathbb{R}^{N}\). In this case, we may instead use the slightly larger alphabet \(\mathscr{A}:=\mathscr{ A}_{K,\delta } + i\mathscr{A}_{K+1,\delta }\) for which L = K(K + 1). This choice indeed corresponds to the one made in [3]. Another natural possibility in this case is to use the 1-norm in \(\mathbb{R}^{2}\) coupled with the diamond lattice as shown in Figure 1 for K = 2.
Note that for the square (or diamond) lattice quantizer of K 2 levels, using the Euclidean norm | ⋅ | on \(\mathbb{C}\) would be sub-optimal. Indeed, the largest value of γ that can be used in (27) is \(\gamma = \frac{K} {\sqrt{2}}\delta\).
Hexagonal norm for a tri-level complex alphabet. It is natural to ask if a complex quantization alphabet \(\mathscr{A}\) with fewer than 4 levels can be used in connection with the noise-shaping quantization algorithm of Lemma 2. For L = 3, we may set \(\mathscr{A}\) to be the vertices of an equilateral triangle in \(\mathbb{C}\) centered at the origin. If the Euclidean norm is used, then it is not difficult to prove that the largest value of γ that can be used in (27) is \(\gamma = \frac{2} {\sqrt{3}}\delta\) (see Figure 2 for a demonstration of this covering). In this case, ∥y∥ ∞ ≤ μ yields ∥u∥ ∞ ≤ δ.
An alternative we have found useful is to employ the norm | ⋅ |∗ induced by a regular hexagonal body whose sides are aligned with the sides of the triangle. Then, as shown in Figure 2, we can attain \(\gamma = \frac{4} {3}\delta\). By choosing the scale of the hexagonal body suitably, we can ensure \(\vert z\vert _{{\ast}}\leq \vert z\vert \leq \frac{2} {\sqrt{3}}\vert z\vert _{{\ast}}\) so that ∥y∥ ∞ ≤ μ implies | y n |∗ ≤ μ for all n, and therefore Lemma 2 yields \(\|u\|_{\infty }\leq \frac{2} {\sqrt{3}}\delta\). Despite the increase in the bound for ∥u∥ ∞ , there is a sizable gain in the “expansion factor” γ∕δ from \(\frac{2} {\sqrt{3}}\) to \(\frac{4} {3}\). This gain is crucial for beta encoding because any β up to this expansion factor is admissible for stability via Lemma 2 provided \(\mathscr{A}\), γ, and δ are suitably scaled to meet (27) and (28) simultaneously.
Rights and permissions
Copyright information
© 2017 Springer International Publishing AG
About this chapter
Cite this chapter
Chou, E., Güntürk, C.S. (2017). Distributed Noise-Shaping Quantization: II. Classical Frames. In: Balan, R., Benedetto, J., Czaja, W., Dellatorre, M., Okoudjou, K. (eds) Excursions in Harmonic Analysis, Volume 5. Applied and Numerical Harmonic Analysis. Birkhäuser, Cham. https://doi.org/10.1007/978-3-319-54711-4_8
Download citation
DOI: https://doi.org/10.1007/978-3-319-54711-4_8
Published:
Publisher Name: Birkhäuser, Cham
Print ISBN: 978-3-319-54710-7
Online ISBN: 978-3-319-54711-4
eBook Packages: Mathematics and StatisticsMathematics and Statistics (R0)