Home > Seminars > Dr. Abla Kammoun
Teaching

Estimation of high dimensional covariance matrices By Dr. Abla Kammoun (KAUST)

  • Class schedule:  Thursday, May 8th, 2014 from 12:00 pm to 01:00 pm
  • Location: Building 9, Lecture Hall I, Room 2322
  • Refreshments:  Pizza and soft drinks will be available @ 11:50 am 
  • Seminar material available at: link

Abstract
In several signal processing applications, the estimation of the covariance matrix of a series of independent multivariate observations is a crucial issue. A reliable estimate is for instance needed in principal component analysis, direction of arrival and estimation, blind subspace methods, and capacity estimation. In the case where the dimension of the observations N is small compared to the number n of observations, the empirical covariance matrix of observations provides a good estimate for the unknown covariance matrix. This estimate becomes however much less accurate when the dimension N gets higher. An interesting theoretical framework for modern estimation occurs when the number of observations scales with the dimension. Recently, several attempts have been done to address this issue, using the theory of large random matrices. The resulting estimation techniques are consistent in the sense that their error tends to zero as N and n grow to infinity at the same pace. The aim of this presentation is to provide an overview of the most recent uses of these techniques in the field of wireless communication.

Biography
Abla Kammoun was born in Sfax, Tunisia. She received the Diplome D'Ingenieur from the Ecole Polytechnique de Tunisie, La Marsa, Tunisia, and the Master's degree and the Ph.D. degree in digital communications from Telecom Paris Tech, France. From June 2010 to April 2012, she was a Postdoctoral Researcher in the TSI Department, Telecom Paris Tech and then at Supélec as part of the Alcatel-Lucent Chair on Flexible Radio until December 2013. She is currently a Postodoctoral Fellow at KAUST. Her research interests include performance analysis, random matrix theory, and semi-blind channel estimation.