EC533: Digital Signal Processing

Lecture 1• Introduction• Real-Time DSP Systems

EC533: Digital Signal Processing

• Instructor: Dr. Mohamed El‐MahallawyContact info. :    E‐mail: mahallawy@ieee.org

Room: 309Office Hours: Wednesday 3rd

• Teaching Assistant: Eng. El‐Nasser S. YusefContact info. :    E‐mail: nasseryousef6@yahoo.com

Room: 314Office Hours:  Tuesday 3rd, 4th.

• References:– Text Book: Digital Signal Processing, A Practical Approach, E. C. 

Ifeachor & B. W. Jervis, 2nd Edition, Prentice Hall, 2002.– Reference: Applied Signal Processing, Concepts, Circuits, and Systems, 

N. Hamdy, CRC Press, 2009.

Assessment System

Assessment Tool Marks

4th week quiz 2.5

6th week quiz 2.5

Lab reports (up to 7th week ) 5

7th week Lab quiz 5

7th week exam 15

11th week quiz 5

Lab reports (up to 12th week ) 5

12th week exam 10

Lab project 5

Final Lab exam 5

Final exam 40

Total 100

Course Outline• Introduction.• Real‐Time DSP Systems.• Discrete‐Time Signals & Systems.• Characteristics of Discrete‐Time Signals.• Z‐Transform.• Digital Filter Structure.• Finite Impulse Response (FIR) Filter Design.• Infinite Impulse Response (IIR) Filter Design.• Discrete Fourier Transform (DFT) & Fast  Fourier Transform (FFT). 

1- Introduction

A)Digital : Signals are either Analogue, Discrete, or Digital signals.

Text Book : Chapter 1, Sections: 1.1, 1.2.

• Analogue Signal :Continuous in both time and amplitude, any value at any time can be found.

• Discrete Signal :Discrete in time (sampled signal) & Continuous in amplitude.

• Digital Signal :Discrete in time (sampled signal) & Discrete in amplitude (Quantized Samples).

1.1 – What is Digital Signal Processing ?

00.

22 0.44 0.

64 0.82 0.

98 1.11 1.

2 1.24

1.27

1.24

1.2

1.11

0.98

0.82

0.64

0.44

0.22

0-0

.22

-0.4

4-0

.64

-0.8

2-0

.98

-1.1

1-1

.2-1

.26

-1.2

8-1

.26

-1.2

-1.1

1-0

.98

-0.8

2 -0.6

4 -0.4

4 -0.2

20

-2

-1.5

-1

-0.5

0

0.5

1

1.5

2

1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37

B) Signal : It is an information-bearing function, It is either:1-D signal as speech.

2-D signal as grey-scale image {i(x,y)}.

3-D signal as video {r(x,y,t),g(x,y,t),b(x,y,t)}.

C) Processing : Signal Processing refers to the work of manipulating signals so that information carried can be expressed, transmitted, restored,… etc in a more efficient & reliable way by the system (hardware \ software).

Least resource usage

Least error

• General Purpose Processors (GPP), Micro‐Controllers.• Digital Signal Processors (DSP); Dedicated Integrated Circuits.

• Programmable Logic (PLD, FPGA).

Fast

Faster

Real‐time DSP’ing

• Programming Languages: Pascal, C, C++,...• High‐Level Languages: Matlab, MathCad,…• Dedicated Tools  (e.g. Filter design s/w packages).

• Greater FlexibilityThe same DSP hardware can be programmed and reprogrammed to perform a variety of functions.

• Guaranteed PrecisionAccuracy is only determined by the number of bits used. (not on resistors,…etc; analogue parameters).

• No drift in performance with temperature or age.• Perfect ReproducibilityIdentical Performance from unit to unit is obtained since there are no variations due to component tolerance. e.g. a digital recording can be copied or reproduced several times with the same quality.

• Superior PerformancePerforming tasks that are not possible with ASP, e.g. linear phase response and complex adaptive filtering algorithms.

• DSP benefits from the tremendous advances in semiconductor technology. Achieving greater reliability, lower cost, smaller size, lower power consumption, and higher speed.

1.2 – Why DSP ?

• Speed & Cost Limitations of ADC & DACEither too expensive or don’t have sufficient resolution for large-bandwidth DSP applications.

• Finite Word-Length ProblemsDegradation in system performance may result due to the usage of a limited number of bits for economic considerations.

• Design TimeDSP system design requires a knowledgeable DSP engineer possessing necessary software resources to accomplish a design in a reasonable time.

1.3 – DSP LIMITATIONS

What is DSP Used For?

……And much more!And much more!

Application Areas

Image Processing              Instrumentation/Control         Speech/Audio MilitaryPattern recognition           spectrum analysis                  speech recognition       secure communicationsRobotic vision  noise reduction                          speech synthesis  radar processingImage enhancement         data compression                      text to speech                 sonar processingFacsimile position and rate control          digital audio           missile guidanceanimation  equalization

Telecommunications        Biomedical Consumer applicationsEcho cancellation patient monitoring cellular mobile phonesAdaptive equalization scanners UMTS (universal Mobile Telec. Sys.)ADPCM trans‐coders EEG brain mappers digital television Spread spectrum ECG Analysis digital camerasVideo conferencing X‐Ray storage/enhancement internet phone 

etc.

DSP Devices & Architectures

• Selecting a DSP – several choices:– Fixed‐point;

– Floating point;

– Application‐specific devices(e.g. FFT processors, speech recognizers,etc.).

• Main DSP Manufacturers:– Texas Instruments (http://www.ti.com)

– Motorola (http://www.motorola.com)

– Analog Devices (http://www.analog.com)

2.1 – Typical Real-Time DSP System

1st Image Frequency

2.2 – Sampling Theorem & Aliasing

Time Domain Frequency Domain

2.2 – Sampling Theorem & Aliasing - continued

• In practice, aliasing is always present because of noise & the existence of signal outside the band of interest.• The problem then is deciding the level of aliasing that is acceptable and then designing a suitable anti-aliasing filter & choosing an appropriate sampling frequency to achieve this.

2.3 – Anti-aliasing Filtering

To reduce the effect of aliasing:a)Sharp Cut-off anti-aliasing filters are normally used to band-limit the signal.b)Increasing the sampling frequency to widen the separation between the signal & the image spectra.c)Practical LPF provides sufficient attenuation at f > fN ; f > fstop to a level not detectable be ADC,

where n is the no of bits used by ADC

dBnA n

76.102.6)25.1log(20min

+=×=

2.3.1 – Butterworth(LPF)

2.3.1 – Butterworth(LPF) - continued

Higher Nnarrower transition width (steeper roll-off).more phase distortion.allows the use of low sampling rate.slower, cheaper ADC

Higher fsfast, expensive ADC. (real-time signal

processing trend).usage of a simple anti-aliasing filter which

minimizes phase distortion.Improved SNR.

See illustrative examples in book, P. 45 54 on how to choose the sampling frequency & aliasing control.