ADSP-01-EC623-ADSP by harshi446


More Info
EC623: Advanced Topics in DSP
    3-0-2-8 Credit Elective Course
              S. R. M. Prasanna

                Dept of ECE,
                IIT Guwahati,

                                     EC623: Advanced Topics in DSP (ADSP) – p. 1/1
Course Planning for Jan-Apr 2008
 Broad Topics
    Multirate Signal Processing (MSP)
    Time Frequency Analysis (TFA)
    Audio Coding (AC): Application of MSP & TFA
 Theory Lectures
   MSP: 14
   TFA: 14
   AC: 14
   Total: 42

                                         EC623: Advanced Topics in DSP (ADSP) – p. 2/1
     Course Planning (contd.)
  MSP: 03
  TFA: 03
  AC: 03
  Total: 09
Written Assignments
  MSP: 02
  TFA: 02
  AC: 02
  Total: 06

                          EC623: Advanced Topics in DSP (ADSP) – p. 3/1
     Time-Table & Evaluation
   3 Lectures / week
     1 L for MSP (Tuesday, 9-10)
     1 L for TFA (Wednesday, 9-10)
     1 L for AC (Friday, 12-13)
   Lab: Wednesday, 15-17
  Quiz: 20 Marks (10%) (End of Jan)
  Mid-Sem: 40 Marks (25%) (End of Feb)
  End-Sem: 60 Marks (40%) (End of Apr)
  Laboratory & Assignments: Continuous (25%)
  Total: 100%

                                      EC623: Advanced Topics in DSP (ADSP) – p. 4/1
Overview of Signal Processing Area
 Relation among different subjects
   Signals & Systems is the foundation part of signal
   processing area
   DSP is the central part of signal processing area
   ADSP is the peripheral part of signal processing

                                           EC623: Advanced Topics in DSP (ADSP) – p. 5/1
             Signals & Systems
Classification of signals
Properties of systems
LTI System
Correlation & Convolution
Differential / Difference equation representation
Fourier representation: CTFS, CTFT, DTFS and DTFT
Concepts of impulse response, frequency response,
system function & transfer function

                                            EC623: Advanced Topics in DSP (ADSP) – p. 6/1
Sampling theorem: time and frequency domains
DFT and its properties
Circular shift and convolution
FFT methods for DFT computation
Analysis and design of IIR and FIR filters
Structures for realization of filters
Finite word length effects

                                            EC623: Advanced Topics in DSP (ADSP) – p. 7/1
Any advanced topic on DSP
Multirate DSP
Time-frequency analysis
Statistical signal processing
Architecture of DSP processors
Speech, Audio, Image and Video processing
Biomedical signal processing
Signal processing for communication

                                       EC623: Advanced Topics in DSP (ADSP) – p. 8/1
      Focus of EC 623 Course
Multirate DSP (MSP)
Time-frequency analysis (TFA)
Audio coding (AC): Applications of MSP & TFA

                                        EC623: Advanced Topics in DSP (ADSP) – p. 9/1
            Topics from MSP
Introduction and Applications of MSP.
Decimation, Interpolation and Fractional Rate.
Polyphase Representation for MSP.
Introduction to Filter Banks.
Maximally Decimated Filter Banks.
Two QMF and M Channel Filter Banks.
Polyphase Representation of Filter Banks.
Perfect Reconstruction (PR) and Paraunitary PR.
Two and M Channel Paraunitary Filter Banks.
Linear Phase PR QMF Banks.
Cosine Modulated Filter Banks and Pseudo QMF Bank.
Recent trends in the Design of Filter Banks.
                                            EC623: Advanced Topics in DSP (ADSP) – p. 10/1
            Topics from TFA
Introduction and Applications of TFA.
Time and Frequency Description of Signals.
Instantaneous Frequency and the Complex Signal.
The Uncertainty Principle.
Densities and Characteristic Functions.
Time-Frequency Distributions: Fundamental Ideas.
TFA using STFT and Wavelets.
Wigner and Wigner-Ville Distribution.
General and Kernel Method for designing TFDs.
TFDs based on Kernel Methods.
Comparison of Various TFDs.
Recent Trends in TFA.
                                          EC623: Advanced Topics in DSP (ADSP) – p. 11/1
             Topics from AC
Introduction to Audio Coding.
Psychoacoustic Principles for Audio Coding.
Filter Banks for Audio Coding.
Transform Coders for Audio Coding.
Subband Coders for Audio Coding.
Sinusoidal Coders for Audio Coding.
Audio Coding Standards and Algorithms.
Lossless Audio Coding.
Quality Measures for Perceptual Audio Coding.
Recent Trends in Audio Coding.

                                         EC623: Advanced Topics in DSP (ADSP) – p. 12/1
      Books (Text/ References)
P. P. Vaidyanathan, "Multirate systems and filter banks",
Pearson, Delhi, 2004.
L. Cohen, "Time-Frequency analysis", Prentice hall, NJ,
USA, 1995.
B. Boashash, "Time-frequency signal analysis and
processing" Elsivier, 2003.
A. Spanias, et. al., "Audio signal processing and
coding", Wiley-Interscience, NJ, USA, 2007.

                                          EC623: Advanced Topics in DSP (ADSP) – p. 13/1
    Unirate v/s Multirate DSP
Unirate: Processing of signals at one sampling rate.
Multirate: Processing at multiple sampling rates.
Merits of Multirate DSP.
  Reduced complexity & distortion.
     All digital implementation.
     No distortion due to ADC & DAC.
  Sampling rate conversion.
  Reduced transmission rate / storage requirement.
     On an average reduced number of bits / sample.
     Subband coding.

                                           EC623: Advanced Topics in DSP (ADSP) – p. 14/1
 Applications of Multirate DSP
Theory of filter banks and their application.
Signal compression: subband coding.
All digital sampling rate conversion.

                                           EC623: Advanced Topics in DSP (ADSP) – p. 15/1
     Time-Frequency Analysis
Time Domain Analysis:
   Amplitude v/s time
   Localization in time domain
   Good resolution in time domain
   Poor resolution in frequency domain
Frequency Domain Analysis:
   Amplitude v/s frequency
   Localization in frequency domain
   Good resolution in frequency domain
   Poor resolution in time domain

                                         EC623: Advanced Topics in DSP (ADSP) – p. 16/1
Time-Frequency Analysis (contd.)
Time-Frequency Analysis:
   Amplitude v/s time and frequency.
   Localization in time and frequency domains.
   Ideally good resolution in both the domains.
   Simultaneous good resolution is not possible.
   Good resolution in one domain with a compromise
   on resolution in other domain.

                                       EC623: Advanced Topics in DSP (ADSP) – p. 17/1
Tools for Time-Frequency Analysis
 Short Term Fourier Transform (STFT).
 Time-frequency distributions (TFDs).

                                        EC623: Advanced Topics in DSP (ADSP) – p. 18/1
              Audio Coding
Audio: Signals in AF (20 Hz - 20 kHz) range.
Speech and music signals.
Speech: 20 Hz - 8 kHz with max energy up to 4 kHz.
Music: 20 Hz - 20 kHz.
Audio processing: Mainly processing music signals.
Audio coding: Different audio compression methods.
Finds immense application in entertainment industry.
MSP and TFA concepts are being extensively used in
audio coding methods.

                                         EC623: Advanced Topics in DSP (ADSP) – p. 19/1

To top