Adaptive Signal Processing Project
Projektkurs i adaptiv signalbehandling
Course description, spring 2004
Lecture Schedule, spring 2004
Course homepage and results, 2004
This project-oriented course is aimed to broaden, reinforce and deepen
the knowledge about advanced
signal processing algorithms and applications.
A series of lectures provides a theoretical
background on adaptation algorithms,
tracking, model-based filter design and inverse filtering.
projects provide opportunities to test and evaluate advanced algorithms
on challenging problems, mainly using acoustic signals.
They also provide experience of working within
a project group.
The projects all manipulate sound in different challenging
applications, using adaptive signal processing techniques.
Signals and Systems, Signal Processing. Furthermore,
and Coding will be a prerequisite for
projects on acoustic digital communication.
System identification. On-line adaptation of models
of linear dynamic systems.
Room acoustics and multipath propagation.
Adaptive filter structures: FIR, IIR,
and filter banks.
Adaptation and tracking algorithms:
RLS and Kalman-based adaptation.
Hypermodel-based design of Kalman-based
trackers and of algorithms with constant gains.
LMS and RLS revisited. Analysis of adaptation
speed and convergence. Frequency-domain adaptation.
Effects of bad excitation and non-stationary signals.
Implementation: Practical aspects,
numerical effects, fixed-point
calculations, the use of signal processors.
Groups of 3-4 students can choose among projects aimed at:
- Active noise control in ventilation ducts
and computer chassis.
- Acoustic Digital Communication with adaptive TDMA
- Acoustic Digital Communication with adaptive OFDM
- Inversion of loudspeaker dynamics and room acoustics
using direct adaptation
Blind acoustic source separation using nonlinear
The projects will utilize fast PC's and signal processors.
A miniproject focuses on adaptation theory.
Written report and oral presentation
of the main project, and a demonstration of the resulting system. Each
group will also develop their own webpage,
which presents the results.
Handouts and research papers.
- Course responsibility:
Department of Material Science,
Signals and Systems Group.