Plenary Lecture

Robust Algorithms of Signal Processing

Professor Tõnu Trump
Department of Radio and Telecommunication Engineering
Tallinn University of Technology

Abstract: This talk concentrates on using the robust statistics in developing signal processing algorithms. Robust statistics continue to gain importance because of increasing amount of environments where noise is impulsive, not Gaussian. Impulsive noise can be seen as the reason of having outliers in the data and signal processing algorithms designed for Gaussian noise cannot cope with the impulsive noise well. The plenary talk breafly reviews the common teqhniques for estimation and detection designed for Gaussian noise. We then highlight the weeknesses of the ordinary technuques when used in the environment with impulsive noise. After that we introduce robust statistics and discuss several robust techniques for estimation and detection.

Brief Biography of the Speaker: Tõnu Trump received his Ph.D. degree from Tallinn University of Technology in 1993. He was from 1994 to 2006 with Ericsson AB in Stockholm Sweden, where he reached the position of expert in echo cancellation and voice enhancement devices. From 2002 to 2006 he was also the rapporteur of Question 17, Study Group 16 at International Telecommunication Union (ITU-T) in Geneva Switzerland. Since 2006 he has been the professor of Signal Processing at Tallinn University of Technology in Tallinn, Estonia.
Prof. Trump has published a number of scientific papers and is the author of more than 10 patents.