The Laboratory of Model-Driven Engineering for Embedded Real-Time Systems (LISE), which is a part of the CEA LIST (450 researchers in the field of software-intensive systems, see http://www-list.cea.fr/), and DIGITEO partner, has an open position for a PhD Student in the area of model-driven engineering and distributed real-time, embedded systems.
CONTEXT:
LISE offers a creative international and highly competitive research environment. The lab is responsible, for research activities related to methods and tools for the development of real-time embedded systems. In this scope, one of the prevailing themes of the laboratory is the application of model-driven approaches to real-time embedded systems development. The lab has a recognized expertise in this field, both on the academic and industrial side. It participates in numerous European and national research projects (Artist2, Atesst, OpenEmbeDD,…), has steady and close relations with some major actors of the domain, and is strongly involved in international standardisation organizations like the Object Management Group.
THESIS SCOPE:
As the size and complexity of embedded systems increase, the specification and design of the overall architecture is becoming ever more important. In order to be able to make the right system architecture selection, the feasibility of candidate application-architecture bindings need to be predicted w.r.t. different functional and non-functional requirements and constraints. Design space exploration and sensitivity analysis techniques assist designers to efficiently decide among candidate alternatives especially when the space of possible solutions is large. The decisions are usually located according to several design goals, and the alternatives therefore represent a multi-criteria decision problem. In an integrated MDD process of embedded systems, these criteria are typically distributed in different model views which provide specific non-functional constraints (deadlines, resource capacity, memory size) and guarantees (predictions) obtained from various information sources (performance simulation, scheduling analysis, reliability analysis, etc.). The task of integrating this heterogeneous information in a global design decision process is indeed difficult [4].
The focus of the proposed thesis is to research on sound means to integrate non-functional information from different validation and verification viewpoints and provide an optimized global solution to a given design decision problem. The thrust of this research has to be put in the specification mechanisms for integrating multidimensional information, its treatment (transformation/refinement) to enable mathematical calculation, and the methodological basis to allow developers for a systematic use of the proposed approach. The development of new or refined mathematical algorithms (search strategies) to explore design alternatives by optimizing objective functions is not a core subject in this thesis, as far as numerous and useful results can be found in the literature [ 1, 2]. However, the suitability of existing algorithms for complex and heterogeneous embedded systems has to be evaluated. The proposed solution must be supported in a MDA approach using UML as the backbone specification language. In particular, the UML profile for MARTE should be used as supporting domain specific modeling language.
It is expected that the results of this thesis will have an important impact on the use of non-functional verification techniques in industrial environments. Indeed, in spite of their potential (“theoretical”) benefits, the current use of formal verification techniques is far from the industrial level. Indeed, their use is still a craft activity, usually reserved to a very limited number of experts. Mathematical results usually only consider specific (not to say limited) behavioral patterns and are limited to tackle small-size systems. However, compositional methods are being studied to support modular analysis techniques and to provide global predictions for the whole system [3].
To summarize, this thesis must enable the design architect of embedded system to find all possible design trade-offs, and analyze them in order to select the most suitable solution according to the various requirements and constraints.
APPLICATION
The position is funded for a period of 3 year (CTBU status: further information about salaries may be found in French language at http://www-instn.cea.fr/rubrique.php3?id_rubrique=67).
The candidate must hold a master in Computer Science or equivalent. He should have a background in model-driven engineering technologies (UML, Eclipse-based tools), good mathematical skills, and knowledge in embedded real-time systems (not mandatory). English is mandatory, French is a plus, but is not required.
Applications shall be sent by e-mail to “sebastien dot gerard at cea dot fr”
or postal mail to
Sébastien Gérard
CEA Saclay
91191 Gif-sur-Yvette cedex
France
Applications have to contain a short CV with background information and marks of master courses.
Best regards,
REFERENCES :
<LI style="MARGIN: 0cm 0cm 0pt">Künzli, Simon, “Efficient design space exploration for embedded systems”, PhD Thesis, Eidgenössische Technische Hochschule ETH Zürich, April 2006. <LI style="MARGIN: 0cm 0cm 0pt">Razvan Racu, Arne Hamann, Rolf Ernst, Bren Mochocki, Xiaobo Sharon Hu: “Methods for power optimization in distributed embedded systems with real-time requirements”. CASES 2006: p. 379-388 <LI style="MARGIN: 0cm 0cm 0pt">Marek Jersak. “Compositional Performance Analysis for Complex”, Embedded Applications. PhD thesis, Technical University of Braunschweig, 2004.
- Huáscar Espinoza, “An Integrated Model-Driven Framework for Specifying and Analyzing Non-Functional Properties of Real-Time Systems”, PhD Thesis, University of Evry, FRANCE. September 2007.
--
Huascar ESPINOZA
CEA SACLAY
DRT/DTSI/SOL/LISE
Tel: +33 1 69 08 45 87
Fax: +33 1 69 08 20 82
91191 GIF/YVETTE CEDEX
FRANCE |
登录
