Advanced Topics in High-Order Statistics

ATHOS - 6620

Work Area: Adaptive Signal Processing and Automatic Control

Keywords high-order statistics, signal and image processing, automatic control, fast algorithms

Start Date: 1 September 92 / Duration: 30 months / Status: running

[ participants / contact ]

Abstract High-order statistics (HOS) may be seen as emerging techniques in the areas of signal processing and automatic control. ATHOS addresses theoretical questions related to HOS, and also investigates their performance in various applications, as well as specific algorithms and architectures.


High-order statistics (HOS) have received increasing interest for several years because of their ability to address some problems that have remained unsolved using conventional techniques. However, due to their recent emergence and their often high computational load, very few algorithms developed for this purpose are used in operational systems. Theoretical topics as well as applications to real world problems are investigated in this working group.


Activities of ATHOS include: visits of researchers or student researchers for durations ranging from one week to two months; yearly meetings; participation in workshops on HOS; participation in relevant international conferences, the organisation of courses on HOS; and the organisation of international workshops open to external participants. An extension of the cooperation to American colleagues is expected with the help of the European Commission and the National Science Foundation.

Progress and Results

The credibility of the advanced algorithms developed by some of the partners is increased after use by others, and makes them more accessible to a wider audience. Confidence in HOS-based techniques is also improved through courses given to the non-specialist scientific community. A better understanding of the HOS potential is expected by partners from the feedback received from other institutions, whether involved in the working group or not. Since industrial applications are already known and under study, it is hoped that HOS advanced techniques will be effectively utilised in operational systems in the longer term.


Since September 1992, ATHOS members have written more than 50 papers related to High-Order Statistics and their applications. In 1993 ATHOS organised two meetings focused on the Estimation and Computation of Cumulants (London, 20-21 May 1993) and System identification and High-Order Statistics (Sophia-Antipolis, Nice, France, 20-21 September 1993). ATHOS has also organised a course on High-Order Statistics for Signal Processing at London on 17-19 May, 1993. Proceedings will be available in 1994).

ATHOS plans to organise two similar meetings in 1994. ATHOS is in charge of organising the fourth International Signal Processing Workshop on High-Order Statistics in June 1995 in Spain, hopefully in cooperation with the IEEE.


Thomson Sintra ASM - F
Parc Sophia Antipolis
BP 138


Katholieke Universiteit Leuven - B
Ruhr Universität Bochum - D
Universität Erlangen-Nürnberg - D
Universidad Politecnica de Catalunya - E
Ecole Normale Supérieure de Lyon - F
Electricité de France - F
Telecom Paris - F
Thomson CSF-SDL - F
Université de Nice-Sophia Antipolis - F
Université de Rouen - F
University of Athens - GR
Università di Perugia - I
Università di Roma La Sapienza - I
Università di Trieste - I
Imperial College of Science, Technology and Medicine - UK
Defence Research Agency - UK


Dr. P. Comon
tel +33/92 96 42 44
fax +33/92 96 39 77

LTR synopses home page LTR work area index LTR acronym index LTR number index LTR Working Groups index
All synopses home page all acronyms index all numbers index

ATHOS - 6620, August 1994

please address enquiries to the ESPRIT Information Desk

html version of synopsis by Nick Cook