PhD Thesis: Smart Modeling
Context
“Software is eating the world!” [1], with massive digitalization of entire business markets (e.g., travel/music/video/photo/book industry…) and the development of cyber-physical systems (CPS) which assist citizens and companies in their daily life and businesses (e.g., smart cities/farming/building, industry 4.0 and transportation systems). But, may the world drive software? Modern CPS require more than ever to integrate a proper understanding of the world in which they involve, in order to both identify the relevant digital innovations the systems may address, and to take informed and open decisions during their dynamic adaptations that act on the physical world. Since the modern CPSs drive our daily life and businesses, they play a political role that requires an understanding of the world far beyond the direct environment captured through the utilization of monitoring techniques and sensors. It also requires to care about the economic, social, and environmental context in which the CPS is involved, in order to take into account all the possible externalities on society raised by the digital innovations and the use of smart CPS. While software engineers are now well equipped for designing safe and deterministic solutions to well-defined problems, they currently face the difficulty to integrate more global concerns considering the underlying wicked problems of the design and use of such systems.
Various disciplines use models for different purposes. Engineers, e.g., software engineers, use prescriptive models to represent the system to implement, and scientists, e.g., environmentalists, use descriptive models to represent the complexity of the world to understand and reason over it for analysis purpose. While the former tries to integrate all the properties in between the various engineering involved in the development process, the latter use models to internalize all the possible externalities of any changes, and later perform trade-off analysis. With the advent of smart CPS, the combination of descriptive and prescriptive models becomes essential, respectively for openly and freely involving massive open data, descriptive and predictive models in the decision process (either for trade-off analysis or dynamic adaptation purposes), and prescriptive models to support the smart design and reconfiguration process of modern CPS. It urges to provide the relevant facilities to software engineers for integrating into the future CPS the various models existing from the scientific community, and thus to support informed decisions, a broader engagement of the various stakeholders (e.g., scientists, decision makers and the general public), and dynamic adaptations with regards to the expected political impact of the smart CPS.
Objective
Modeling environments (incl., Integrated Development Environments, seen as special cases for the so-called programming languages) take very different forms depending on the users, the modeling objectives, and the contexts in which the modeling activity fits. During the last decade these environments have evolved to integrate ever more advanced features, and offering the user abstractions and automation (e.g., code generation, V&V and DevOps). Generative approaches currently offer the means to automatically support modeling languages (i.e., Domain-Specific Language) with environments offering facilities for editing, analysis and debugging; all offering support for the implementation of an already well-designed system. However, we are lacking in facilities supporting the system design activity itself (tools to support thinking leading to an optimal design of a desired system). These socio-technical systems can be software, cyber-physical, or physical systems.
Description of the work
The research program aims to establish the required foundations to integrate not only prescriptive engineering models, but also data (from the systems or existing ones, measurements, user preferences, requirements, data from the surrounding environment, etc.), descriptive models of the environment (physical, but also environmental, economic and social), and predictive models to offer the recommendations and predictions required to assist the modeling activities.
The main objectives are:
- the definition of a unified modeling framework making it possible to integrate heterogeneous models, and to establish the relationships between them.
- the definition of a language engineering approach allowing to design and integrate heterogeneous modeling languages, but also the associated modeling environments:
- in connection with distributed computing to offer modular, distributed and collaborative environments
- in connection with programming languages to offer a sound combination of live programming, literate programming and exploratory programming.
- in connection with model reduction techniques, approximate computing, and statistical approaches to offer recommendation and prediction tools.
This program is grounded into the application domain of cyber-physical systems in order to offer design space exploration tools to design ground breaking products, and to finely monitor existing systems for predictive maintenance (e.g., digital twin). The program is also grounded in the domain of scientific computing, with the main objective to explore tools that ease the integration of heterogeneous scientific models, and the interaction with them (i.e., tools for tradeoff analysis and decision making through “what if” scenarios).
The long term objective is to cross-fertilize the aforementioned application domains, with the overall objective to develop smarter cyber-physical systems in the future. The candidates will leverage ongoing collaborations to investigate sustainability system, i.e., smart cyber-physical systems dedicated to sustain the production, transport and consumption of ressources. These systems include smart cities, farming, building, transportation and grids. The transition to more sustainable production, exploitation and consumption patterns requires technological, scientific and social breakthroughs. This transition must meet human needs while ensuring the preservation of natural resources. This requires communication between several disciplines of the environmental sciences, natural sciences and social sciences in the time scale to measure the degree of irreversibility of our current and future choices, but also in the spatial scale to determine the level of adaptation of the solutions envisaged in a given environment. On the one hand, this requires developing scientific knowledge and methods (eg, green chemistry, renewable energies) as well as socio-economic policies (eg, fiscal and redistribution policy) on our capacity for the sustainable use of our resources, but also in the adaptive capacity of ecosystems. On the other hand, it is necessary to determine the future evolution of needs and their conditions of access in order to prevent various forms of injustice, sources of conflicts and wars.
Environment
The candidate will work at Inria in the DiverSE team. Inria is the French national institute for research in computer science. There are 8 Inria research centres located throughout France, hosting more than 200 research teams. The DiverSE team is located in Rennes. DiverSE’s research is in the area of software engineering. The team is actively involved in European, French and industrial projects and is composed of 9 faculty members, 20 PhD students, 2 post-docs and 4 engineers. The main advisor of the PhD thesis will be Prof. Benoit Combemale (University of Rennes 1, DiverSE team).
How to apply
Please send your application (PDF) as soon as possible. Screening of applications starts immediately and continues until the position is filled. Send a cover letter including names of at least two referees, CV, PDFs of publications (if any) to the main advisor.
References
- https://www.wsj.com/articles/SB10001424053111903480904576512250915629460
- Benoit Combemale, Jörg Kienzle, Gunter Mussbacher, Hyacinth Ali, Daniel Amyot, et al.. A Hitchhiker’s Guide to Model-Driven Engineering for Data-Centric Systems. IEEE Software, Institute of Electrical and Electronics Engineers, 2020, pp.1-9. Preprint: https://hal.inria.fr/hal-02612087
- Gunter Mussbacher, Benoit Combemale, Jörg Kienzle, Silvia Abrahão, Hyacinth Ali, et al.. Opportunities in Intelligent Modeling Assistance. Software and Systems Modeling, Springer Verlag, 2020, pp.1-7. Preprint: https://hal.inria.fr/hal-02876536
