Skip to Main content Skip to Navigation

Partitionnement en ligne d'applications flots de données pour des architectures temps réel auto-adaptatives

Abstract : The current challenges of the development of the complex
embedded systems such as the integrated systems of image
processing, consist to successfully realizing products reliable,
powerful, inexpensive and effective whatever the conditions of use. To take up
these challenges passes by a good choice of architecture, methods and tools
adapted to the applications concerned and target technologies. For many
applications, in particular in telecommunication and multi-media, soft real time
realizations are often sufficient, i.e. implementations aiming to obtaining a
quality of service adapted to the needs. Instead of being based on the Worse
Case Execution Times (WCET) or sequences of test often not very
representative to design these systems, our approach targets a self-adapting
platform able to be configured during the execution of the application (thus on
line). We can quote as examples of applications the case of a fixed camera of
remote monitoring which adapts its processing according to the nature of the
acquired images or a multimode mobile terminal which changes its standard of
transmission if the quality of the communication channel requires it.
The reconfigurable components have levels of performances and flexibility
which make them very attractive in a growing number of developments.
Dynamic reconfiguration (partial or full) makes it possible to re-use the same
hardware resources for a succession of processing, and this in similar way to a
software realization. We propose an approach allowing to allocate and
schedule dynamically the tasks of a data flow application according to an
estimate of their execution times in order to respect the time constraints as well
as possible. This on line reconfiguration requires research of compromise
complexity/effectiveness of the allocation and scheduling in order to optimize
the quality of service and to reduce their costs of realization.
Complete list of metadata

Cited literature [153 references]  Display  Hide  Download
Contributor : Fakhreddine Ghaffari <>
Submitted on : Wednesday, January 31, 2007 - 1:23:39 PM
Last modification on : Wednesday, October 14, 2020 - 4:23:56 AM
Long-term archiving on: : Wednesday, April 7, 2010 - 2:35:37 AM


  • HAL Id : tel-00128242, version 1



Fakhreddine Ghaffari. Partitionnement en ligne d'applications flots de données pour des architectures temps réel auto-adaptatives. Micro et nanotechnologies/Microélectronique. Université Nice Sophia Antipolis, 2006. Français. ⟨tel-00128242⟩



Record views


Files downloads