Avis de Soutenance
Monsieur Luca GRECO
Automatique
Soutiendra publiquement ses travaux d’habilitation à diriger les recherches
« Resource-aware control »
Soutenance prévue le lundi 7 octobre 2024 à 9h30
Lieu : Salle sc.071 – Bâtiment Bouygues
CentraleSupélec 9 Rue Joliot-Curie 91192 Gif-sur-Yvette
Composition du jury Mme. Elena PANTELEY Directrice de Recherches CNRS, L2S Rapporteuse
Mme. Maria Domenica DI BENEDETTO Professeure, DEIE – Université de L’Aquila Rapporteuse
M. Luca ZACCARIAN DR CNRS à LAAS et Professeur à l’Université de Trento Rapporteur
M. Mohammed CHADLI Professeur, IBISC Université Paris-Saclay Examinateur
M. Constantin MORARESCU Professeur, CRAN Université de Lorraine Examinateur
Abstract:
The past decades have witnessed a dramatic growth in interest for complex distributed interconnected systems and their control.
The common denominator of the aforementioned systems is the interaction between physical objects and abstract digital entities that we devise in order to sense, monitor and control the former ones.
These systems come with a significant level of complexity and heterogeneity that can be very hard to deal with. To aid in this daunting task, it is beneficial to adopt a high level point of view based on the distinction between platform and control applications. The platform is an abstraction layer providing control applications with resources such as communication channels and computation capabilities and imposing realistic constraints. The services guaranteed by the platform can be characterised in terms of a quantifiable Quality of Service (QoS) (e.g. bandwidth, delays, scheduling guarantees, fairness constraints). On the other hand, control applications exploit the resources provided by the platform to accomplish their tasks with a desirable Quality of Control (QoC) (e.g. stability, robustness, performance). Resource-aware control is all about the study of this deep relation between QoS and QoC, between platform and control applications. The goal is to investigate how platform limitations impact the QoC and conversely how control strategies can be designed to achieve some desired QoC if they are aware of the platform’s QoS.
The QoS received by the control applications can widely change in time (even with a short time scale). In order to cope with this inherent variability, we pursue two dual research paths: flexible control and flexible platform. The former aims at developing control laws whose structure or characteristic parameters can dynamically change to adapt to the varying QoS, always guaranteeing a desired QoC. The latter aims at developing flexible policies to manage platform parameters (e.g. computation time allotted to control tasks by the operating system scheduler, time slots in the access to a communication channel) to ensure a desired QoC.
Finally, we will illustrate some future research perspectives. The first is related to resource-aware control. The other concerns a new research topic in the field of neurosciences and aims to understand the basic mechanisms behind an intriguing phenomenon called Self-Organised Criticality (SOC).
