Number of hours
- Lectures 14.0
- Projects 0
- Tutorials 4.0
- Internship 0
- Laboratory works 8.0
ECTS
ECTS 2.0
Goal(s)
Give basis on mean square linear filtering for the random signals. Methods description with adaptive experimental approaches
Contact Florent CHATELAINContent(s)
- 1 : Wiener filtering in the continuous representation
- Non-causal Wiener filter
- Causal Wiener filtering: Bode and Shannon approach
- 2 : Discrete Wiener filtering with finite memory
- FIR Wiener filter
- optimal linear prediction and autoregressive models
- 3 : Adaptive algorithms
- 4 : Discrete Kalman filtering
Prerequisites
- Basic signal processing course
- Random signals and spectral analysis course
- notions of power spectral density, Wiener-Khintchine theorem
- non-parametric spectral estimation: periodogram
- Notions of optimization: quadratic optimization, gradient descent algorithm
Test
First Session 1
- ET1: Written exam (2 hours)
- CC1: Lab preparation + lab reports
Second Session
- ET2: Written exam (2 hours)
- CC1 = CC2 (no make-up possible for continuous assessment)
Exam conditions: One A4 sheet (double-sided, handwritten, personal) is allowed during the written exam (both sessions 1 and 2).
Session 1 : N1 = 50% examen1 + 50% CC1
Session 2 : N2 = 50% examen2 + 50% CC1 (repris de la session 1)
Bibliography
- Detection, Estimation and Modulation Theory, Part 1, Harry L. VAN TREES Wiley, 1968
- Optimal Filtering, Brian D. O. Anderson and John B. Moore. Dover Publications, 2005