PROCESSAMENTO DIGITAL DE SINAIS E FILTROS DIGITAIS

TEORIA DE PROCESSAMENTO DIGITAL DE SINAIS E TEORIA DE FILTROS DIGITAIS

1. AGC
DSP
Professor A G Constantinides© 1
Digital Signal Processing &
Digital Filters
An Introductory Course
By
Professor A G Constantinides
MSc, EE4, ISE4, PhD

2. AGC
DSP
Professor A G Constantinides© 2
Digital Signal Processing &
Digital Filters
Contents
1-Introduction
1) Introduction to Digital Signal Processing
Review of background DSP
Review of mathematical methods
Review of discrete-time random
processes and linear systems

3. AGC
DSP
Professor A G Constantinides© 3
Digital Signal Processing &
Digital Filters
2)Multirate techniques and wavelets
Introduction to short-time Fourier analysis
Filter-banks and overlap-add methods of analysis
and synthesis
Introduction to generalised time-frequency
representation
Wavelet analysis
Multirate signal processing
Interpolation and decimation
Efficient filter structures for interpolation and
decimation

4. AGC
DSP
Professor A G Constantinides© 4
Digital Signal Processing &
Digital Filters
3)Classical spectrum estimation methods
Power spectrum, power spectral density functions,
random processes and linear systems
Introduction to statistical estimation and estimators
Biased and unbiased estimators
Einstein/Wiener Khintchine Theorem
Estimation of autocorrelations
Means and variances of periodograms
Smoothed spectral estimates, leakage

5. AGC
DSP
Professor A G Constantinides© 5
Digital Signal Processing &
Digital Filters
4)Modern spectrum estimation methods
Introduction to modern spectral estimation:
Principles and approaches
Cramer-Rao Lower Bound (CRLB) and Efficient
estimators
The Maximum Entropy Method (MEM) or
Autoregressive Power Spectrum Estimation:
Principles.
The MEM equations and Levinson/Durbin
algorithm

6. AGC
DSP
Professor A G Constantinides© 6
Digital Signal Processing &
Digital Filters
4)Modern spectrum estimation methods
(continued)
Introduction to Linear Prediction
Linear Predictive Coding using covariances
and correlations
Cholesky decomposition
Lattice Filters
Linear Prediction of Speech Signals

7. AGC
DSP
Professor A G Constantinides© 7
Digital Signal Processing &
Digital Filters
5)Adaptive signal processing
Introduction to adaptive signal processing
Objective measures of goodness
Least squares and consequences
Steepest descent
The LMS and RLS algorithms
Kalman Filters

8. AGC
DSP
Professor A G Constantinides© 8
Digital Signal Processing &
Digital Filters
6)Applications
Communications
Biomedical
Seismic
Audio/Music

9. AGC
DSP
Professor A G Constantinides© 9
DIGITAL FILTERS
Digital Filters
In this course you will learn:
How to choose an appropriate filter
response.
Why Butterworth responses are maximally
flat.
Why Chebyshev and Elliptic responses are
equiripple.
When to choose an IIR and when an FIR
filter

10. AGC
DSP
Professor A G Constantinides© 10
DIGITAL FILTERS
How do you design FIR and IIR filters from
specifications on amplitude performance?
What are multirate systems and their
properties? What is interpolation /
Upsampling and Decimation / Downsampling?
How do you design efficient Decimation and
Interpolation systems?
What are frequency transformations and how
do you design these?
How accurate is the DFT as a spectrum
estimator?

11. AGC
DSP
Professor A G Constantinides© 11
DIGITAL FILTERS
What are short FFT algorithms?
How do you choose the required
wordlength?
What are Fast Convolutions and how
are they realised?
How do you deal with a DSP problem in
practice?

12. AGC
DSP
Professor A G Constantinides© 12
Course content
Assumed DSP background
DSP Backgroundfolder
1-Introduction
2-z transform
3-transfer functions
4-Signal Flow Graphs
5-digital filters intro

13. AGC
DSP
Professor A G Constantinides© 13
Course content
2-Digital Filter Design
1-Digital Filters (FIR)
2-Digital Filters (IIR)
3-Multirate
1-Interpolation_Decimation

14. AGC
DSP
Professor A G Constantinides© 14
Course content
4-Tranforms
1-DFT
2-DFT_one2two
3-general transforms
4-Wavelets
5-Finite Wordlength
1-Finite Wordlength

15. AGC
DSP
Professor A G Constantinides© 15
Course content
6-Spectrum Estimation(Assumed
background inMathematical
Backgroundfolder)
1-Fourier transform & DFT
2-FFT-based Power Spectrum Estimation
3-Modern Spectrum Estimation
4-Intro-Estimation
5-Eigen-based methods
6-A Prediction Problem

16. AGC
DSP
Professor A G Constantinides© 16
Course content
7-Adaptive Signal Processing
1-Adaptive Signal Processing
8-Applications
1-Applications
2-Applications

17. AGC
DSP
Professor A G Constantinides© 17
Digital Signal Processing &
Digital Filters
BOOKS
Main Course text books:Digital Signal
Processing: A computer Based Approach, S K
Mitra, McGraw Hill
Mathematical Methods and Algorithms for
Signal Processing, Todd Moon, Addison
Wesley
Other books:
Digital Signal Processing, Roberts & Mullis,
Addison Wesley
Digital Filters, Antoniou, McGraw Hill

18. AGC
DSP
Professor A G Constantinides© 18
DIGITAL FILTERS
Analogue Vs Digital Signal Processing
Reliability:
Analogue system performance degrades due to:
Long term drift (ageing)
Short term drift (temperature?)
Sensitivity to voltage instability.
Batch-to-Batch component variation.
High discrete component count
Interconnection failures

19. AGC
DSP
Professor A G Constantinides© 19
DIGITAL FILTERS
Digital Systems:
No short or long term drifts.
Relative immunity to minor power supply
variations.
Virtually identical components.
IC’s have > 15 year lifetime
Development costs
System changes at design/development
stage only software changes.
Digital system simulation is realistic.

20. AGC
DSP
Professor A G Constantinides© 20
DIGITAL FILTERS
Power aspects
Size
Dissipation
DSP chips available as well as ASIC/FPGA
realisations

21. AGC
DSP
Professor A G Constantinides© 21
Applications
Radar systems & Sonar systems
 Doppler filters.
 Clutter Suppression.
 Matched filters.
 Target tracking.
Identification

22. AGC
DSP
Professor A G Constantinides© 22
DIGITAL FILTERS
Image Processing
Image data compression.
Image filtering.
Image enhancement.
Spectral Analysis.
Scene Analysis / Pattern recognition.

23. AGC
DSP
Professor A G Constantinides© 23
DIGITAL FILTERS
Biomedical Signal Analysis
Spatial image enhancement. (X-rays)
Spectral Analysis.
3-D reconstruction from projections.
Digital filtering and Data compression.

24. AGC
DSP
Professor A G Constantinides© 24
DIGITAL FILTERS
Music
Music recording.
Multi-track “mixing”.
CD and DAT.
Filtering / Synthesis / Special effects.

25. AGC
DSP
Professor A G Constantinides© 25
DIGITAL FILTERS
Seismic Signal Analysis
Bandpass Filtering for S/N
improvement.
Predictive deconvolution to extract
reverberation characteristics.
Optimal filtering. (Wiener and Kalman.)

26. AGC
DSP
Professor A G Constantinides© 26
DIGITAL FILTERS
Telecommunications and Consumer
Products
These are the largest and most
pervasive applications of DSP and
Digital Filtering
Mobile Communications
Digital Recording
Digital Cameras
Blue Tooth or similar