Advanced DSP methods

Course guarantor:

Lecturer:

Tutor:

Supervisor:

Department of Circuit Theory

Synopsis:

The course introduces advanced methods of analysis and processing of digital signals such as correlation, spectral, coherence or cepstral analysis, as well as methods of decomposition into principal and independent components, methods for determining the relationship between random signals and basic classification techniques used in signal analysis. Attention is paid to practical applications of the mentioned techniques, e.g. for noise suppression or compression.

Requirements:

Knowledge of basic techniques of digital signal processing, digital filtering as well as mathematical apparatus for describing continuous and discrete signals and systems is assumed.

Syllabus of lectures:

1. LPC analysis: calculation of AR model parameters, LPC spectrum

2. General signal modeling (AR, MA, ARMA)

3. Delay measurement using correlation and spectral analysis

4. Coherence function, magnitude square coherence (MSC) and its application

5. Cepstral analysis and its application

6. Spectral and cepstral distance and their application

7. Reduction of additive and convolutional noise in the spectral and cepstral domains

8. Discrete cosine transform

9. Principal component analysis (PCA) as a basis for lossy signal compression

10. Basics of classification (k-means, GMM, SVM)

11. Use of neural networks in signal processing

12. Implementation of discrete wavelet transform by filter bank, quadrature filters

13. Principles of blind separation and deconvolution methods of signals

14. Reserve

Syllabus of tutorials:

1. LPC analysis, LPC spectrum

2. Signal modeling (AR, MA models of the 1st and 2nd order)

3. Delay measurement based on cross-power spectral density

4. Properties and applications of the coherence function

5. Real and complex cepstrum - definition and basic properties

6. Cepstral distance

7. Suppression of additive noise in the frequency domain

8. Calculation and use of the discrete cosine transform

9. Principal component analysis and KLT transform

10. Classification based on k-means

11. Classification based on GMM

12. Noise suppression based on ANN

13. Wavelet transform, implementation by a filter bank, noise suppression based on WT

14. Reserve

Study Objective:

Students will learn to use the above-mentioned advanced signal analysis techniques, interpret the results obtained, and practically use basic classification techniques.

Study materials:

[1] Oppenheim, A. V., Schaffer, R. W. : Discrete-Time Signal Processing. Prentice-Hall, 3rd edition, 2009.

[2] S. V. Vaseghi: Advanced Digital Signal Processing and Noise Reduction, Wiley, 2009.

[3] M. Hayes: Statistical digital signal processing and modeling. Wiley, 1999.

Note:

Further information:

No time-table has been prepared for this course

The course is a part of the following study plans:

• Communications and Internet of Things - Internet of Things (compulsory elective course)
• Communications and Internet of Things - Intelligent Communication Network (compulsory elective course)
• Communications and Internet of Things - Wireless Technology and Photonics (compulsory elective course)
• Communications and Internet of Things - Audiovisual Technology (compulsory elective course)
• Communications and Internet of Things - Communication and Information Processing (compulsory elective course)
• Electronics and Integrated Systems (compulsory elective course)