Studies on the Asymptotic Behavior of Parameters in Optimal Scalar Quantization.
Yee, Victoria B.
2010
Abstract
The goal in digital device design is to achieve high performance at low cost, and to pursue this goal, all components of the device must be designed accordingly. A principal component common in digital devices is the quantizer, and frequently used is the minimum mean-squared error (MSE) or emph{optimal}, fixed-rate scalar quantizer. In this thesis, we focus on aids to the design of such quantizers. For an exponential source with variance $sigma^2$, we estimate the largest finite quantization threshold by providing upper and lower bounds which are functions of the number of quantization levels $N$. The upper bound is $3sigmalog N$, $Ngeq1$, and the lower bound is $3sigmalog N + o_Nleft(1right)sigma-1.46004,sigma$, $N>9$. Using these bounds, we derive an upper bound to the convergence rate of $N^2Dleft(Nright)$ to the Panter-Dite constant, where $Dleft(Nright)$ is the least MSE of any $N$-level scalar quantizer. Furthermore, we present two, very simple, non-iterative and non-recursive suboptimal quantizer design methods for exponential sources that produce quantizers with good MSE performance. For an improved understanding of the half steps and quantization thresholds in optimal quantizers as functions of $N$, we use as inspiration the result by Nitadori~cite{Nitadori1965} where, exploiting a key side effect of the source's memoryless property, he derived an infinite sequence such that for any $N$, the $k$th term of the sequence is equal to the $k$th half step (counting from the right) of the optimal $N$-level quantizer designed for a unit variance exponential source. In our work, using an asymptotic version of this key side effect which holds for general exponential (GE) sources parameterized by an exponential power $p$ and a utilizing a method of our own devising, we show that for such a source, the $k$th half step of an optimal $N$-level quantizer multiplied by the ($p-1$)st power of the $k$th threshold approaches the $k$th term of the Nitadori sequence as $N$ grows to infinity. Thus, the Nitadori sequence asymptotically characterizes the cells of MMSE quantizers for GE-sources, as well as exponential.Subjects
Quantization Scalar Quantization Optimal Quantization Asymptotic Quantization
Types
Thesis
Metadata
Show full item recordCollections
Showing items related by title, author, creator and subject.
-
Gelca, Răzvan; Uribe, Alejandro (Springer-Verlag; Springer-Verlag Berlin Heidelberg, 2003-02)
-
Lee, Donghoon (1990)
-
Ju, J. D.; Jang, D.; Nwankpa, A.; Atzmon, M. (The American Institute of Physics, 2011-03-01)
Remediation of Harmful Language
The University of Michigan Library aims to describe library materials in a way that respects the people and communities who create, use, and are represented in our collections. Report harmful or offensive language in catalog records, finding aids, or elsewhere in our collections anonymously through our metadata feedback form. More information at Remediation of Harmful Language.
Accessibility
If you are unable to use this file in its current format, please select the Contact Us link and we can modify it to make it more accessible to you.