Signal Processing---ELEN427

A fourth year Elective Course



Lecturer: Dr. A.R.Clark


Course Brief &and Course Outline

Course Objective:

Signal Processing encompasses a wide variety of techniques for performing useful operations on signals and has application in virtually every engineering discipline. Therefore, the course is structured as an introduction to signal processing techniques in common use, mainly using digital signals -- hence Digital Signal Processing. It is important to emphasize the first word Digital. It is difficult, if not impossible, to develop a decent understanding of the topic without getting hands-on experience (ie messing around with computer simulation packages). Thus one of the main course objectives is to familiarize the student with the topic using the tools provided by the matlab software on the student network. MANY of the tutorial problems that have been set relate to exercises using matlab. Note that my objective in setting the tutorial problems is to pass on an understanding of the field of Digital Signal Processing, not the ability to pass examinations. It is useless to attempt the tutorial problems in the week before the examination -- they are not suited to the task.

Prerequisite Knowledge.

Signal Processing draws on linear systems and communications background knowledge, specifically:
Linear Systems Concepts:
Impulse response, convolution.
Laplace Transforms:
Transfer functions, poles, zeros, stability criteria.
Fourier Transforms:
Transfer functions, magnitude and phase response, relations to the Laplace transform.
Sampled Signals:
Sampling theorem - Shannon, Nyquist. Aliasing, Quantization, reconstruction of digital signals.

Prescribed Text

Oppenheim A.V and Shafer R.W Discrete-Time Signal Processing Prentice-Hall 1989

Useful Texts:

Almost any book with the generic title .... Digital Signal Processing .... will probably be of benefit (Several are available in the Engineering library).

Course Content:

Further Linear Systems Concepts.
Review of Linear Systems; group delay; specification of filters in time and frequency domain; approximation concepts of maximally flat and equiripple responses; specification using magnitude squared functions; std filters - Butterworth, Chebyshev, Elliptic etc; filter transformation to high, band etc pass.
Digital Filters.
Review of sampling concepts, aliasing etc; what is a digital filter?; frequency response of digital filter, Laplace transform of digital signal; z-transform \& properties; description of digital filter using the z-transform; Design of Finite Impulse response Digital filters - Impulse Invariant, Fourier Series with Windows, Frequency Sampling methods; Design of Infinite Impulse Response Digital Filters - Impulse Invariant, Bilinear z-transform.
The Discrete Fourier Transform.
Review of the Fourier transform - continuous and discrete in time and frequency domains; Properties of the DFT; the practical DFT - the fast fourier transform, of FFT; Cost of computation - decimation in time, frequency algorithms; Inverse transform; problems with the calculation of the DFT - aliasing, leakage, picket-fencing; the selection of FFT parameters.
Signal Analysis using the DFT.
Discrete Parseval's Theorem; Power Spectral Density; Cross Power Spectrum; Auto & Cross Correlation; Convolution; Basic Averaging; Coherence.
Windows.
Standard Windows - Rect, Hanning, Hamming, Blackman, Kaiser.

Laboratory Experiment

This involves the design of a digital filter to meet individually prescribed specifications. Unfortunately, the experiment will most probably be performed ahead of the lecture material, so self-study is essential prior to doing the laboratory. The filter designed by the student will be downloaded to a DSP56001 signal processing system, and the filter's performance is to be checked on an analogue to analogue basis. The laboratory will count 10% of the total mark of the course. The lab will probably run for a period of two, or perhaps three weeks only. Notices will appear on the 4th year notice board and students will have to book a time slot convenient to them. Laboratory reports are to be handed within one week of performing the experiment at reception. Late reports, and non existent reports will be penalized. It must be remembered that marks are allocated according to the content and style of the written report only - it must of an engineering standard.

Course Project

Each student is required to complete a project and submit a report on a topic to be assigned. Again the report must be of engineering standard, and counts 30% of the total mark of the course. The project hand-in date will be co-ordinated so as not to clash with the other course projects. The project will have a three week duration. The student is expected to spend no more than 20 hours on the project.

Examination

The examination will be of type 2 -- ie Bring in any calculator, and One A4 crib sheet. The exam will count 60% towards the final course mark. The examination will be structured in such a way as give absolutely no advantage to the fancy calculators. Again, I am interested in methods employed, and the understanding/reasoning behind them, and not ``one step ...''

Tutorials

Tutorial questions will be handed out during the course. The fifth period on a Thursday will occasionally be used as a tut, as well as an odd lecture or two, at my discretion. Again, it is important to note that Digital Signal Processing is only really significant in the context of a computer based system. For this reason many of the tut problems are geared towards using matlab. These tut problems are meant to be done at the relevant time during the course to practically highlight the concepts taught. Doing them at the end of the course will simply be a waste of time! You may obviously use any package you like, MathCad etc springs to mind. I shall be using matlab however. It is a common lament that the tut problems are geared towards matlab and that matlab is not provided in the examination. See comment in Course Objective!

Appointments

I have what I call a Modified Open Door policy. I have a high regard for the peer support system - it is often only when you are trying to teach something to your fellows that it really sinks in. Hence when those resources have been tapped, students are to see me as a group at any time. Appointments can be made through the secretary, but I prefer casual drop- ins. Perhaps the best method is to send me a descriptive note on e-mail (CLARK) - and follow up on my reply. However, if you have not been attending lectures or tuts, do not expect me to be at your beck and call. This is especially true before the exam!!

Lectures

Although the Senate has ruled that attendance at lectures is not compulsory, various announcements about course content, laboratory, project and tutorial schedules will be made in lectures.
Since 19 February 1999.