book

Practical Digital Signal Processing

by Edmund Lai

October 2003

Intermediate to advanced

304 pages

7h 46m

English

Newnes

Read now

Unlock full access

1.1 Benefits of processing signals digitally1.2 Definition of some terms1.3 DSP systems1.4 Some application areas1.5 Objectives and overview of the book
2.1 A typical DSP system2.2 Sampling2.3 Quantization2.4 Analog-to-digital converters2.5 Analog reconstruction2.6 Digital-to-analog converters2.7 To probe further
3.1 Notation3.2 Typical discrete-time signals3.3 Operations on discrete-time signals3.4 Classification of systems3.5 The concept of convolution3.6 Autocorrelation and cross-correlation of sequences
4.1 Discrete Fourier series for discrete-time periodic signals4.2 Discrete Fourier transform for discrete-time aperiodic signals4.3 The inverse discrete Fourier transform and its computation4.4 Properties of the DFT4.5 The fast Fourier transform4.6 Practical implementation issues4.7 Computation of convolution using DFT4.8 Frequency ranges of some natural and man-made signals

5.1 Periodic signal generation using wave tables5.2 Wireless transmitter implementation5.3 Speech synthesis5.4 Image enhancement5.5 Active noise control5.6 To probe further
6.1 Classification of digital filters6.2 Filter design process6.3 Characteristics of FIR filters6.4 Window method6.5 Frequency sampling method6.6 Parks-McClelland method6.7 Linear programming method6.8 Design examples6.9 To probe further
7.1 Characteristics of IIR filters7.2 Review of classical analog filter7.3 IIR filters from analog filters7.4 Direct design methods7.5 FIR vs IIR7.6 To probe further
8.1 Direct form8.2 Cascade form8.3 Parallel form8.4 Other structures8.5 Software implementation8.6 Representation of numbers8.7 Finite word-length effects
9.1 Common features9.2 Hardware architecture9.3 Special instructions and addressing modes9.4 General purpose microprocessors for DSP9.5 Choosing a processor9.6 To probe further
10.1 DSP system design flow10.2 Development tools

Content preview from Practical Digital Signal Processing

Appendix A

Binary encoding of quantization levels

Consider an n-bit binary number representing a full-scale range R. In other words, the range R is being quantized into 2ⁿ quantization levels. If R is unipolar, the quantized value x_Q lies in the range [0,R]. If it is bipolar, x_Q lies in the range [−R/2,R/2].

We shall denote the n-bit pattern as a vector b = [b_n-1, b_n-2, …, b₁, b₀] where b_n-1 is called the most significant bit (MSB) and b₀ is the least significant bit (LSB). There are many ways in which this n-bit pattern can be used to encode x_Q. Three most common ways are: