15863 articles – 31997 references  [version française]
HAL: hal-00706772, version 1

Detailed view  Export this paper
SCAN'12, Novosibirsk : Russian Federation (2012)
Interval-based Robustness of Linear Parametrized Filters
Alexandre Chapoutot 1, Thibault Hilaire 2, Philippe Chevrel 3, 4
For the PEPS CNRS ReSyST collaboration(s)
(2012-09-29)

This article deals with the resilient implementation of parametrized linear filters (or controllers), i.e. realizations that are robust with respect to their implementation with fixed-point arithmetic. The implementation of a linear filter/controller in an embedded device is a difficult task because the numerical version of such algorithms suff ers from a deterioration in performances and characteristics. This degradation has two separate origins, corresponding to the quantization of the embedded coefficients and the round-off occurring during the computations. The optimal realization problem is to find, for a given filter, the most resilient realization. We here consider linear filters that depends on a set of parameters that are not exactly known during the design. They are used for example in automotive control, where a very late re-tuning is required. The paper presents results on FWL resiliency analyzis using interval optimization methods [2], and compare them to those obtained with the sensitivity approach.
1:  Unité d'Électronique et d'informatique (UEI)
ENSTA ParisTech
2:  Laboratoire d'Informatique de Paris 6 (LIP6)
CNRS : UMR7606 – Université Pierre et Marie Curie [UPMC] - Paris VI
3:  Institut de Recherche en Communications et en Cybernétique de Nantes (IRCCyN)
CNRS : UMR6597 – PRES Université Nantes Angers Le Mans [UNAM] – École Centrale de Nantes – École Nationale Supérieure des Mines - Nantes – Ecole Polytechnique de l'Université de Nantes
4:  École nationale supérieure des Mines de Nantes
École Nationale Supérieure des Mines - Nantes – Groupe des Écoles des Mines (GEM)
Computer Science/Automatic Control Engineering