Skip to Main content Skip to Navigation

Variability-intensive applications over highly-configurable platforms : Early feasibility and optimality analysis

Abstract : Software-intensive embedded systems, such as automotive systems, are increasingly built from highly- variable applications targeting evermore configurable hardware platforms. Moreover, besides variability present at specification level, there are often various ways to implement a given application on a specific platform. This threefold variability leads to an immense number of system design alternatives. The notorious problem is establishing, at the early stages of development, which designs fulfill and optimize functional and non-functional requirements. Traditional system design frameworks capture system requirements and specifications to derive and evaluate every design automatically. They evaluate every facet of the problem, but they use enumeration et iterative based techniques. Consequently, they may offer poor scalability at both modelling and analysis stages. On the other hand, variability modelling approaches exploit commonalities between different but related products to efficiently evaluate the whole product line. However, given system specifications, they lack to automatically derive the design space while only specific facets of the problem are evaluated in isolation. We propose a model-driven framework that combines and extends both approaches. It captures requirements and specifications in the form of variable data-flows and configurable hardware platforms, with non-functional constraints and a cost function. An original mapping algorithm then derives and automatically generates the design space in the form of a variability-aware model of computation. This formal model encodes every system designs efficiently. We then propose a novel verification algorithm to pinpoint suitable designs efficiently with respect to functional and non-functional requirements. The benefits of our approach are evaluated through a real-world case study from the automotive industry.
Document type :
Complete list of metadata
Contributor : Abes Star :  Contact
Submitted on : Wednesday, April 14, 2021 - 12:02:08 PM
Last modification on : Thursday, April 15, 2021 - 3:20:27 AM
Long-term archiving on: : Thursday, July 15, 2021 - 6:23:31 PM


Version validated by the jury (STAR)


  • HAL Id : tel-03197885, version 1



Sami Lazreg. Variability-intensive applications over highly-configurable platforms : Early feasibility and optimality analysis. Embedded Systems. Université Côte d'Azur, 2020. English. ⟨NNT : 2020COAZ4070⟩. ⟨tel-03197885⟩



Record views


Files downloads