Communication Systems Principles Using MATLAB

5Quantization and Coding

5.1 Chapter Objectives

On completion of this chapter, the reader should

1. Be conversant with the principles of scalar quantization and be able to explain the operation of a vector quantizer.
2. Be conversant with the principles of minimum‐redundancy codeword assignment and understand the important algorithm classes for lossless source coding.
3. Be able to explain several image compression approaches, including the DCT.
4. Understand the basic approach to waveform and parametric speech encoding and be able to explain the advantages and disadvantages of each.
5. Be able to explain the key requirements for audio encoders and the building blocks that go to make up an audio encoding system.

5.2 Introduction

Quantization is the process of assigning a digital value, usually an integer, to represent one or more analog values. Since there are only a certain number of bits allocated to each of these samples, a corresponding number of discrete levels exist. Thus, we can only represent the true analog signal at its nearest approximation. Careful choice of the representation means that we can get by with as few bits as possible.

In addition to the number of representation levels, it is important to have sufficient density of sampling – either samples per second in the case of digitized audio or spatial density in the case of digitized images. The overall data rate (in bits per second or bps) for a sampled audio signal is then the number of samples per second, multiplied ...