ENT773: Introduction to DSP
| School | Cardiff School of Engineering |
| Department Code | ENGIN |
| Module Code | ENT773 |
| External Subject Code | H610 |
| Number of Credits | 10 |
| Level | L7 |
| Language of Delivery | English |
| Module Leader | Dr Mehrdad Ghassempoory |
| Semester | Autumn Semester |
| Academic Year | 2012/3 |
Outline Description of Module
- To provide an introduction to the analysis, synthesis and application of linear digital processing systems for filtering and spectral analysis.
- To provide familiarisation with the MATLAB/Simulink simulation environment for modelling and assessing DSP systems.
On completion of the module a student should be able to
- Comprehend the analysis and design of frequency-selective filters, matched filters, correlators and various types of discrete Fourier transform
- Evaluate the effectiveness of DSP in a wide variety of system applications
How the module will be delivered
16 one-hour lectures
6 one-hour problem classes
- backed up by a website accessible via BLACKBOARD
Lectures and problem classes are carefully scheduled to support the Coursework Assignments.
Students are expected to undertake substantial private study, of which at least 20 hours should be devoted to the two Coursework Assignments, and 6 hours to preparation for the 6 problem classes.
The Coursework Assignments will provide a useful framework for revision for the examination.
6 one-hour problem classes
- backed up by a website accessible via BLACKBOARD
Lectures and problem classes are carefully scheduled to support the Coursework Assignments.
Students are expected to undertake substantial private study, of which at least 20 hours should be devoted to the two Coursework Assignments, and 6 hours to preparation for the 6 problem classes.
The Coursework Assignments will provide a useful framework for revision for the examination.
Skills that will be practised and developed
- Model linear DSP systems in terms of recurrence relations and Z-transforms
- Perform digital filter design to meet a prescribed specification
- Use a variety of techniques and software tools to design linear DSP systems
- Employ MATLAB/Simulink skills to design, for example: telecommunication, radar, sonar, biomedical and control systems
How the module will be assessed
This module is assessed using two components:
1. two coursework assignments (40%)
2. a 2-hour written examination (60%)
In order to pass the module and obtain 10 credits, the minimum pass mark of 50% must be achieved in items 1 and 2 combined.
1. two coursework assignments (40%)
2. a 2-hour written examination (60%)
In order to pass the module and obtain 10 credits, the minimum pass mark of 50% must be achieved in items 1 and 2 combined.
Assessment Breakdown
| Type | % | Title | Duration(hrs) |
|---|---|---|---|
| Exam - Autumn Semester | 60 | Introduction To Dsp | 2 |
| Written Assessment | 40 | Coursework | N/A |
Syllabus content
Revision of FIR and IIR digital frequency-selective filter analysis and design
Filter architectures, canonical forms. Time domain descriptions: system equations, impulse response. Frequency domain descriptions: Z-transform, frequency response (magnitude, phase, group delay), pole-zero plots, 3D visualisation of Z-domain. FIR designs via MATLAB’s fir1.m and Remez’ algorithm: lowpass, highpass, bandpass, bandstop, linear phase conditions. IIR designs: Butterworth, Chebyshev and Elliptic: nonlinear phase characteristics. Allpass IIR filters with nonlinear phase, use in phase linearisation of IIRs. Approximate FIR-IIR equivalence. Examples of digital filtering applications.
Matched filtering
Convolution, correlation and the matched filter condition. Optimal detection using a matched filter or serial correlator. Performance criteria: correlation peaks and sidelobes, peak-to-sidelobe ratio. Detection theory, optimal threshold design. Binary m-sequence properties, generation and detection - use in CDMA applications. Linear FM chirp waveform - use in pulse compression radar. Bandpass chirps. Impact on performance of coefficient rounding: dichotomisation. Examples of matched filter applications.
Discrete Fourier transformation
The slow Fourier transform (SFT) The correlator bank SFT. The filterbank approach to spectral analysis. Fast Fourier transform (FFT) techniques: decimation in time and frequency, in-place algorithms, bit-reversed inputs and outputs; signal flowgraphs based on butterfly units and twiddle factors. Reduction in computational complexity. Numerical artifacts in the DFT spectrum: sampling, aliasing and spectral leakage for single-tone and multi-tone signals. Advantages and disadvantages of signal windowing. Accurate calculation of Fourier series using the FFT. The spectrogram. The Chirp-Z transform, the windowed CZT, CZT as SFT. Examples of DFT applications.
Filter architectures, canonical forms. Time domain descriptions: system equations, impulse response. Frequency domain descriptions: Z-transform, frequency response (magnitude, phase, group delay), pole-zero plots, 3D visualisation of Z-domain. FIR designs via MATLAB’s fir1.m and Remez’ algorithm: lowpass, highpass, bandpass, bandstop, linear phase conditions. IIR designs: Butterworth, Chebyshev and Elliptic: nonlinear phase characteristics. Allpass IIR filters with nonlinear phase, use in phase linearisation of IIRs. Approximate FIR-IIR equivalence. Examples of digital filtering applications.
Matched filtering
Convolution, correlation and the matched filter condition. Optimal detection using a matched filter or serial correlator. Performance criteria: correlation peaks and sidelobes, peak-to-sidelobe ratio. Detection theory, optimal threshold design. Binary m-sequence properties, generation and detection - use in CDMA applications. Linear FM chirp waveform - use in pulse compression radar. Bandpass chirps. Impact on performance of coefficient rounding: dichotomisation. Examples of matched filter applications.
Discrete Fourier transformation
The slow Fourier transform (SFT) The correlator bank SFT. The filterbank approach to spectral analysis. Fast Fourier transform (FFT) techniques: decimation in time and frequency, in-place algorithms, bit-reversed inputs and outputs; signal flowgraphs based on butterfly units and twiddle factors. Reduction in computational complexity. Numerical artifacts in the DFT spectrum: sampling, aliasing and spectral leakage for single-tone and multi-tone signals. Advantages and disadvantages of signal windowing. Accurate calculation of Fourier series using the FFT. The spectrogram. The Chirp-Z transform, the windowed CZT, CZT as SFT. Examples of DFT applications.
Essential Reading and Resource List
Getting started with MATLAB, R. Pratap, (Oxford University Press 2002)
Digital Signal Processing using MATLAB, V. K. Ingle and J. G. Proakis, (Brooks/Cole 2000)
Digital Signal Processing : A practical approach, E. C. Ifeachor and B. W. Jervis, (Addison Wesley 2nd Edition 2002)
Digital Signal Processing : Principles, algorithms and applications, J. G. Proakis and D. G. Manolakis, (Brooks/Cole 2002)
Digital Signal Processing : Spectral Computation and Filter Design, C-T. Chen, (Oxford University Press 2001)
”The Chirp-Z Transform Algorithm”, L. R. Rabiner, P. W. Schafer and C. M. Rader, IEEE Trans. Audio and Electroacoustics, Vol. AU-17, No. 1, pp 7-13, March 1968.
Digital Signal Processing using MATLAB, V. K. Ingle and J. G. Proakis, (Brooks/Cole 2000)
Digital Signal Processing : A practical approach, E. C. Ifeachor and B. W. Jervis, (Addison Wesley 2nd Edition 2002)
Digital Signal Processing : Principles, algorithms and applications, J. G. Proakis and D. G. Manolakis, (Brooks/Cole 2002)
Further Reading
Understanding Digital Signal Processing, R. C. Lyons, (Prentice -Hall 2001)Digital Signal Processing : Spectral Computation and Filter Design, C-T. Chen, (Oxford University Press 2001)
”The Chirp-Z Transform Algorithm”, L. R. Rabiner, P. W. Schafer and C. M. Rader, IEEE Trans. Audio and Electroacoustics, Vol. AU-17, No. 1, pp 7-13, March 1968.
Copyright Cardiff University.
Registered charity no. 1136855