Ctrl-A seminars

at the Imag building, in room 406

GLSI Axis Workshop

the research axis of the LIG lab dedicated to Software and Information System Engineering (GLSI, in french) is grouping the teams Vasco, Sigma and Ctrl-A, and having its yearly GLSI Axis Workshop (WAX!) of 2024.

The Ctrl-A team contributed two presentations :

• 09.50 – Ctrl-A : Jolahn Vaudey – DSL & model-based Self-reconfiguration of industrial control systems as a response to cyberattacks
• 11.05 – Ctrl-A : Kouds Halitim – Autonomic HPC resources management with practical combinations of Control and Scheduling.

.

” amphithéâtre=”” bergés-b007″=”” ground=”” floor=”” of=”” ense3,=”” 21=”” av.=”” des=”” martyrs,=”” 38031=”” grenoble
and visio :
Zoom Link
https://univ-grenoble-alpes-fr.zoom.us/j/96923827649?pwd=T0NkUnE3SHlVNiswMzRYTXNvdkdQZz09

ID de réunion: 969 2382 7649
Code secret: 6102246321

Estelle Hotellier
(these CIFRE Naval Group)

Title: “Specification-based Intrusion Detection for Hierarchical Hybrid Industrial Control Systems”

Abstract:
This presentation tackles the question of cybersecurity through network intrusion detection for Industrial Control Systems (ICSs).
Our interest arises from the increasing number of cybersecurity incidents targeting ICSs and the need to detect attacks that manipulate
the physical process. Such attacks are called process aware attacks: they are sophisticated cyberattacks aiming at disrupting the
physical process and inducing incorrect behaviors of the system.

In our work, we propose a specification-based, process aware, Intrusion Detection System (IDS) for ICSs.
Our approach aims to link safety specifications and security properties. Thus, we use international and
industry standards specifications concerning local safety, global safety and networks of the industrial process,
in order to systematically obtain security properties. The obtained security properties are cybersecurity related requirements.
They are translated into security patterns in order to be runtime monitored by our network IDS.
We rely on specification language formalism such as Linear Temporal Logic (LTL), Metric Temporal Logic (MTL)
and Signal Temporal Logic (STL) to express temporal properties and tackle the hybrid dynamics of ICSs.

.

at Imag building, amphitheater

Quentin GUILLOTEAU

(PhD, Ctrl-A/Datamove)

Title :
«Control-based runtime management of HPC systems with support for reproducible experiments»

Abstract:

High-Performance Computing (HPC) systems have become increasingly more complex, and their performance and power consumption make them less predictable.
This unpredictability requires cautious runtime management to guarantee an acceptable Quality-of-Service to the end users.
Such a regulation problem arises in the context of the computing grid middleware CiGri that aims at harvesting the idle computing resources of a set of cluster by injection low priority jobs.
A too aggressive harvesting strategy can lead to the degradation of the performance for all the users of the clusters, while a too shy harvesting will leave resources idle and thus lose computing power.
There is thus a tradeoff between the amount of resources that can be harvested and the resulting degradation of users jobs, which can evolve at runtime based on Service Level Agreements and the current load of the system.

We claim that such regulation challenges can be addressed with tools from Autonomic Computing, and in particular when coupled with Control Theory.
This thesis investigates several regulation problems in the context of CiGri with such tools.
We will focus on regulating the harvesting based on the load of a shared distributed file-system, and improving the overall usage of the computing resources.
We will also evaluate and compare the reusability of the proposed control-based solutions in the context of HPC systems.

The experiments done in this thesis also led us to investigate new tools and techniques to improve the cost and reproducibility of the experiments.
We will present a tool named NixOS-compose able to generate and deploy reproducible distributed software environments.
We will also investigate techniques to reduce the number of machines needed to deploy experiments on grid or cluster middlewares, such as CiGri, while ensuring an acceptable level of realism for the final deployed system.

Jury:

– Alexandru COSTAN (INSA Rennes, IRISA) – Maître de conférence, HDR – Rapporteur
– Alessandro PAPADOPOULOS (Mälardalen University) – Professeur – Rapporteur
– Fabienne BOYER (Université Grenoble Alpes) – Maîtresse de conférence, HDR – Examinatrice
– Georges DA COSTA (Université Paul Sabatier) – Professeur – Examinateur
– Noël DE PALMA (Université Grenoble Alpes) – Professeur – Examinateur
– Eric RUTTEN (Inria Grenoble) – Chargé de recherche, HDR – Directeur de thèse
– Olivier RICHARD (Université Grenoble Alpes) – Maître de conférence – Co-encadrant de thèse

This will be a PhD defense rehearsal : see dec. 11 for abstract

Abstract:
The ever-growing complexity and size of high-performance computing (HPC) systems pose significant challenges, particularly in maintaining efficiency. This study embarks on an exploration of a novel approach grounded in control theory to tackle the dynamic regulation of power in HPC architectures. The core idea revolves around the continuous monitoring of an application progress and the real-time adjustment of power caps for processors, aiming to strike a harmonious balance between energy efficiency and computational intensity characteristic of HPC. This approach utilizes cascaded control strategy, incorporating proportional-integral (PI) control and Model Predictive Control (MPC). It is seamlessly integrated into the established framework of the Argo Node Resource Manager, and its effectiveness is evaluated across several clusters of Grid’5000. These assessments are conducted using a standard compute-bound HPC benchmark intricately combined with Intel’s Running Average Power Limit (RAPL) energy-saving mechanism. Through rigorous algorithmic analysis and an extensive array of computational experiments, this research seeks to bridge the gap between energy efficiency and the relentless computational demands of HPC applications. By harnessing the precision and strategic potential of RAPL-guided control mechanisms, this work aspires to usher in a more sus- tainable and efficient era in the realm of high-performance computing.

Keywords: HPC, Energy consumption, RAPL, System Identification, PI, MPC

joint work with Sophie Cerf at Inria/CRIStAL Spirals Lille,

and also Raphaël Bleuse & Jonathan Bleuzen

.

Lilian Gardon (intern, IUT)

Title :
« Optimisation d’une structure de données pour le calcul et la simulation»

Abstract:

Durant mon stage au sein de l’équipe CTRL-A ma tâche a été d’otpimiser la structure de données ProcSet. Cette structure de données est utilisée dans les logiciels OAR et BatSim qui permettent l’ordonnancement de
tâches sur des plages de processeurs pour les infrastructures hautes performances. Le projet existant est en Python, le but de ce stage et de l’écrire en C, pour pouvoir gérer nous-mêmes la mémoire utilisée par procset.py lors de son utilisation, et diminuer certains coûts d’exécution due à Python. Une des contraintes importante de ce projet est de garder une interface python pour les utilisateurs de ProcSet.

Quentin Guilloteau (PhD UGA)

Title :
« Under Control: a practical introduction of Control Theory for Computing Systems»

Abstract:
remake of the tutorial, proposed to the LIG : see below Fri. 21 apr. 2023

Quentin Guilloteau (PhD, UGA, Lig, Ctrl-A/Datamove)

Title :
« Reusability of Autonomic Controllers in High Performance Computing»

Abstract:

High Performance Computing systems are subject to dynamical variations occurring in e.g., the execution of the jobs, the quantity of I/O, the network consumption, etc. Such systems thus need Autonomic management with online regulation. In our work, we consider High Performance Computing (HPC) systems, and more particularly CiGri, a system harvesting idle resources in a computing grid. It submits jobs from Bag-of-tasks applications with the lowest priority to the clusters in order to maximize their use. This harvesting introduces perturbations for the premium users of the cluster, e.g., degraded performances of the fileserver. Therefore, the submission must be regulated in a feedback loop with a Control Theory approach. The design of such a controller can involve considerable competence and work, and it is highly desirable that it can be as reusable as possible, adaptable to different contexts where it is instantiated. In this work we propose to study the relationship between this reusability and the techniques of adaptive control, by describing an approach involving classical PID control, adaptive PI control, and Model-Free Control (MFC), and performing experimental validation and comparison in the use-case of CiGri, evaluating several criteria and in particular their reusability and adaptivity to variations.

• Rosa Pagano (MSc from U. Milano)mid-internship presentation
• Quentin Guilloteau (PhD UGA)
  Title :
  « Under Control: a practical introduction of Control Theory for Computing Systems»

  Abstract:

  Computing systems are getting more and more complex. The software stacks are growing, and are executed on top of complex hardware. The behavior of applications is thus becoming extremely difficult to predict, or model, correctly. We need regulation to provide Quality-of-Service guarantees.
  One approach is to have a theoretical model of the system and design complex algorithm based on this model. Such an approach is limited by the quality of the model. Modelling all the possible cases, and behaviors is tedious and error-prone. A different approach to regulate the behavior of a computing system is to periodically take measurements of the system metrics, and adapt the input based on these measurements. This approach is called closed-loop, and this is the interest of the Autonomic Computing community. There are several ways to adapt the input of the system. One can use ad-hoc solutions based on arbitrary rules or IA for example. These solutions do not provide any proven guarantees on the closed-loop system. On the other hand, the Control Theory field has been applying math-proven methods to closed-loop physical systems for decades. But the application of Control Theory to computing systems is only quite recent.

  We believe that numerous computing systems or applications could benefit from feedback loops architecture, and that tools from Control Theory provide interesting guarantees compared to other strategies. This tutorial aims at introducing the attendees with Control Theory, and present the methodology through a practical example.

  At the end of the tutorial, attendees will have:

  – Learned about Control Theory motivations and tools
  – Identified a model of a system
  – Designed a Proportional Controller, and understood its limitations
  – Designed a more complex controller: a Proportional-Integral Controller

atelier SEnS : Sciences, ENvironnements, Sociétés

suivant le modèle disponible là

Ghada Moualla (ATER IUT, UGA, Lig)

Title :
« Self-adaptive Device Management for the IoT Using Constraint Solving »

Abstract :
In the context of IoT (Internet of Things), Device Management (DM), i.e., remote administration of IoT devices, becomes essential to keep them connected, updated and secure, thus increasing their lifespan through firmware and configuration updates and security patches. Legacy DM solutions are adequate when dealing with home devices (such as Television set-top boxes) but need to be extended to adapt to new IoT requirements. Indeed, their manual operation by system administrators requires advanced knowledge and skills. Further, the static DM platform — a component above IoT platforms that offers advanced features such as campaign updates / massive operation management — is unable to scale and adapt to IoT dynamicity. To cope with this, this work, performed in an industrial context at Orange, proposes a self-adaptive architecture with runtime horizontal scaling of DM servers, with an autonomic Auto-Scaling Manager, integrating in the loop constraint programming for decision- making, validated with a meaningful industrial use-case.

Ali El Hadi Noura (MISCIT, MSc defense rehearsal)

« Integration of Scheduler Knowledge into CiGri Control Loop, using Feed-Forward Control »

Abstract :
Harvesting resources in data-centers can be done by injecting best-effort jobs, in a way
that minimizes unused resources while not perturbating higher priority jobs.
Controlling this injection has been done using control-theory approaches, in a purely reactive fashion, independently of the scheduler and Resource and Jobs Management System (RJMS) in the system.
This work explores a more pro-active approach, where information from the scheduler is used in a Feed-Forward Control loop, in order to achieve better performance.

Jolahn Vaudey (ENSIMAG/MOSIG, MSc defense rehearsal)

** the presentation will be in french **

« Comparison of Controller Synthesis and Scheduling Techniques for Dynamically Reconfigurable Allocation of Tasks on Computing Resources »

Abstract :
Recently, Heptagon/BZR has been used in a context pertaining to cybersecurity, and the problems solved (mapping tasks to resources) had a lot in common with scheduling problems.
This raised some questions during the peer reviews, but until now, there were no attempts to compare these two methods.
During this internship, the goal is therefore to find a problem formalization such that its instances can be solved by both approaches, and subsequently evaluate their relative performances.

• • Ali El Hadi Noura « Integration of RJMS Knowledge into CiGri Control Loop, using Feed-Forward Control

-0- team picture (the first official one !)

-1- short presentations (5 to 10mn) : MSc “mid”-internship and PhD

• Ahmadreza Ahmadi (M2 MISCIT) « Model Free Control (MFC) approach for the collection of resources in dynamic management of High Performance Computing (HPC)» Abstract :
  I am working on a High Performance Computing (HPC) system hosted at Grenoble Alpes University. The system is not fully used by its main users, and a component named ”CiGri” has been added to inject extra tasks on the system to fully use it. The extra tasks share both the computation resources and the file system for storing intermediate data. The injection of extra tasks is managed with a feedback loop. As the environment is very heterogenous and instable, we proposed to use the Model Free Control (MFC) framework to dynamically manage the utilization of the cluster. The goal is to increase the utilization of the system while avoiding the overload of the file system. We used MFC with the following features: 1) a ‘virtual model’ simple representing the unknown system dynamics, 2) elimination of the need of complex system parameter estimation, and 3) assurance of reduced computational costs. The conducted work explores and adopts MFC using the intelligent PID control law, which assures both robustness and quality reference tracking with respect to time varying system dynamics.
• Nelson Nkawa (M2 CSI) « Threat modeling for industrials systems security. » Abstract :
  Nowadays, the modernization of the world is based on computer systems. Among these, we notice the growth of industrial systems. Industrial systems provide a large number of services including building management, production and transport. In order to ensure the proper functioning of these systems, three basic security objectives (availability, data integrity and confidentiality) have to be guaranteed. As these systems are evolving, a method (STRIDE) developed by Microsoft will make possible to evaluate and identify potential threats. Starting from the security objectives we define primitives of this method (STRIDE) for the main command control equipment in order to automate the creation of tasks and associated threats. This will be able to feed our template from the manufacturers’ catalogues, and the threats from the CVE databases. We present here the automatic generation of a template from a provided configuration.
• Mike da Silva (PhD CEA) « Automation of cybersecurity risk analysis of industrial systems. » Abstract: The Information Technologies (IT) are more and more present inside moderns industrial systems beside the Operational Technologies (OT).
  However, IT and OT have different objectives. IT is aimed at data processing while OT is aimed at driving industrial processes. The new design of future factory, “Smart Factories”, break this view converging IT and OT to an hybrid system cyber-physic (Cyber-physical systems, CPS). This new kind of systems exposes the industrial safety infrastructure to IT risks and requires protections. However, the conventional application of cybersecurity are not suitable for this systems. In fact, objectives of cybersecurity are focused on confidentiality, integrity and availibility unlike industial systems which have the dependability as a priority. It is therefore necessary to develop a specific risk analysis for systems that combine either dependability and cybersecurity. To address this problematic, we begin a PhD thesis on “Safety-security models for caracterization of industrial devices security”.

-2- Estelle Hotellier (PhD Naval Group)

« Specification-based Intrusion Detection for Hybrid Industrial Control Systems »

Abstract :
Industrial Control Systems (ICSs) may be targeted by sophisticated cyberattacks that aim to damage the physical process. Our research consist in developing a behavioral, process-aware, Intrusion Detection System (IDS) for complex ICSs. In our approach, complex ICSs are hierarchical control systems built on top of elementary local loops, including discrete distributed control and supervisory control.
Our approach aims to link safety specifications and security properties. The safety properties are extracted from international and industry standards and then instantiated into security patterns monitored by our network IDS. The monitoring of the security patterns is executed at runtime.
This detection system monitors network traffic between the distributed control and the local loops.

Mots clés :

Mots clés : Model-Free Control, Control for Computing, Resource Harvesting

Mots clés : réseau intelligent, micro-réseau, surveillance, énergies renouvelables, automates

• Sophie Cerf « A Combining Machine Learning and Control Theory »
• Mahyar T. Moghaddam « Survey on Middleware support of self-adaptive systems »

• David N. Donkor, working with Bogdan Robu « A model free control approach in utilizing idle computing grid resources without file server overload » Abstract:The recent feedback control approaches adopted in cluster and file server management of High Performance Computing (HPC) systems all utilises the model-based control approach; these model-based control laws inherits the striking flaws including system modelling errors, complex system parameter estimation, high computational costs. The deterministic and time varying nature of our system thus, CIGRI- a lightweight, scalable and fault tolerant grid system which exploits idle computing resources through its interaction with a batch scheduler software has been studied by recent works. The underlining disadvantages of the model-based scheme were highlighted in their results hence the need to take an alternate approach in this study. In this work, we propose the model free control approach with the following features: 1) a ‘virtual model’ representing the unknown system dynamics,2) Elimination of the need of complex system parameter estimation and 3) assurance of less computational costs. This study first explores and adopts model free control using the intelligent PID control law, which assures both robustness and quality reference tracking in respect to time varying system dynamics. Results from both simulations and real time experiments will be covered and presented in later works to explore the intuitive abilities of this approach
• Ismail Hawila, working with Sophie Cerf & Raphaël Bleuse « A Towards an adaptive control for power regulation in HPC » Abstract:Power in high-performance computing (HPC) systems is becoming an important design issue. By the use of feedback loop, appropriate power cap could be provided for processors at runtime by monitoring application progress. The objective of the work is to develop an adaptive control that is robust to various execution platforms and that will adjust the power in HPC systems while maintaining performance.”
• Hassan Hijazi, working with Mahyar T. Moghaddam « Linking agent-based simulation with sensor networks for architectural design decisions » Abstract:Agent-based approaches are used to model IoT systems and their interaction with the dynamic surrounding environment. In the smart cities context, the environment includes various agents, e.g., humans, cars, and autonomous vehicles. The impact of those agents’ behavior on the monitoring IoT software system should be analyzed. Considering those interactions in the design and adaptation of IoT systems could guarantee proper function and quality of services. In our research, we plan to model the dynamics of a smart city environment using agent-based simulations to input the software architecture models of our IoT system. After performing a literature review, we use the GAMA simulator for environment modeling and CAPS and CupCarbon IoT modeling and simulating frameworks to perform various adaptation needed in the software system’s architecture.
• Nicolas Vallet, working with Stéphane Mocanu & Gwenaël Delaval « Sequential Function Chart security properties checking and enforcement » Abstract:The Sequential Function Chart (SFC) language is widely used in the programming of industrial control systems. Such systems have critical security properties, which need to be insured by the control programs. Our approach is to consider tool support for the checking and enforcement of these properties on SFC programs, using the Heptagon/BZR language and compiler.

Abstract:

Chaotic attractors are solutions of deterministic processes, of which
the topology can be described by templates.
We consider templates of chaotic attractors bounded by a genus–1 torus
described by a linking matrix.
In this presentation, we introduce a novel and unique tool to validate
a linking matrix, to optimize the compactness of the corresponding
template, and to draw this template.
We will first detail the different validation steps of a linking matrix.
We then present an algorithm to build a minimal sequence of crossings
from the linking matrix.
Such a sequence leads to a minimal-height depiction of the template.
We will then present future optimization opportunities for this problem.

Avec l’avènement de l’internet des objets (IdO) qui se base sur des objets hétérogènes, dynamiques et de haute volumétries, des besoins en administration à distance. Il s’agit par exemple de mise à jour de logiciels, de configurations, de résolution de problèmes à distance ainsi que de récolte de données de fonctionnement. Ces opérations d’administration permettent d’assurer une bonne qualité de service et d’expérience pour les utilisateurs. Elles permettent en outre, le déploiement de nouvelles fonctionnalités, de correctifs logiciels, et de mise à jour de sécurité.
Les plateformes industrielles existantes d’administration montrent leurs limites avec des parcs formés d’objets statiques, en termes de capacités et d’environnements, comme les passerelles internet domestiques et décodeurs de flux TV. Ces plateformes sont opérées manuellement par des équipes d’administrateurs systèmes et requièrent une expertise conséquente. Concernant les flottes de l’IdO, l’hétérogénéité se traduit en un ensemble d’équipements ayant des capacités différentes de calcul et de connectivité réseau. La dynamicité concerne les environnements de ces équipements qui varient en termes de services en cours d’exécution, de qualité du lien réseau, de capacité restante de calcul. La volumétrie des objets de d’IdO impose un besoin de passage à l’échelle afin de gérer des milliards d’équipements contrairement aux flottes composée de millions d’équipements aujourd’hui.

Par suite, l’administration de flottes de l’IdO requiert une adaptation constante de ces opérations en termes de nature, de vitesse et de cible. Les approches manuelles existantes ne permettent pas de réaliser ces opérations en prenant en compte les spécificités de l’IdO. Afin d’adresser cette problématique, ce travail de thèse industrielle chez Orange, vise à appliquer le paradigme de l’informatique autonomique au pilotage et la distribution des plateformes d’administration. L’objectif est d’assurer que les besoins en administration des flottes de l’IdO soient automatiquement réalisés, et ce, avec une consommation optimale de ressources de calcul et de réseau, ainsi qu’avec un nombre le moins élevé possible, d’erreurs d’exécution.

Notre proposition s’appuie sur quatre boucles autonomiques coordonnées. Deux d’entre elles sont responsables de l’automatisation du maintien à jour de la flotte d’équipements tandis que les deux autres sont chargées de la régulation de l’utilisation des ressources assurant ainsi un passage à l’échelle vertical et horizontal. Notre proposition est validée au travers de deux prototypes. Le premier sert de démonstrateur de l’utilisabilité de notre approche pour le pilotage d’une plateforme industrielle d’administration de l’IdO (Live Objects d’Orange) qui est utilisée en production. Le deuxième démontre les capacités de passage à l’échelle et de tolérance aux fautes de notre proposition. Il s’appuie sur des technologies à code source ouverts. Les résultats sont encourageants par rapport aux approches existantes (p. ex. Vitesse d’exécution multipliée par deux sans augmentation du taux d’équipements en dysfonctionnement).

Résumé de la thèse en anglais:

With the expansion of Internet of Things (IoT) that relies on heterogeneous; dynamic; and massively deployed devices; Device Management (DM), which consists of firmware update, configuration, troubleshooting and tracking, is required for proper quality of service and user experience, deployment of new functions, bug fixes and distribution of security patches.
Existing Home and IoT industrial DM platforms are already showing their limits with a few static home and IoT devices (e.g., routers, TV Decoders). Currently, these platforms are mainly manually operated by experts such as system administrators, and require extensive knowledge and skills. Heterogeneity implies that devices have diverse compute and network capabilities. Dynamicity translates to variation of devices environments (e.g., network quality, running services, nearby devices). The massive aspect is reflected in fleets composed of billions of devices as opposed to millions currently.
Therefore, IoT device administration requires launching administration operations that assure the well-functioning of device fleets. These operations are to be adapted in terms of nature, speed, target, accordingly to devices current service requirements, computing capabilities and network conditions. Existing manually operated approaches cannot be applied on these massive and diverse devices forming the IoT.
To tackle these issues, our work in an industrial research context, at Orange Labs, proposes applying autonomic computing to platform operation and distribution. It aims to ensure that administration requirements of a device fleet are automatically fulfilled using the optimal amount of resources and with the least amount of execution errors.
Specifically, our contribution relies on four coordinated autonomic loops. The first two loops are responsible for handling fleet variation and update operations dispatching, while the remaining two others focus on vertical and horizontal scalability. Our approach allows automatic administration platform operation, more accurate and faster error diagnosis, vertical and horizontal scaling along with simpler IoT DM platform administration.
For experimental validation, we developed two prototypes: one that demonstrates the usability of our approach with Orange’s industrial IoT platform for its piloting, while the other one demonstrates vertical scalability using extended open-source remote administration software. Our prototypes show encouraging results, such as two times faster firmware upgrade operation execution speed, compared to existing legacy telecommunication operator approaches.

Mahyar Tourchi Moghaddam (post-doc, CPS4EU project)

« IAS: an IoT Architectural Self-adaptation Framework »

Abstract:
This work develops a generic approach to model control loops and their interac- tion within the Internet of Things (IoT) environments. We take advantage of MAPE-K loops to enable architectural self-adaptation. The system’s architectural setting is aligned with the adaptation goals and with the components run-time situation and constraints. We introduce an integrated framework for IoT Architectural Self-adaptation (IAS) were functional control elements are in charge of environmental adaptation and autonomic control elements handle the functional system’s architectural adaptation. A Queuing Networks (QN) approach was used for modeling the IAS. The IAS-QN can model control levels and their interaction to per- form both architectural and environmental adaptations. The IAS-QN was modeled on a smart grid system for the Melle-Longchamp area (France). Our architectural adaptation approach successfully set the propositions to enhance the performance of the electricity transmission system. This research is a part of CPS4EU European industrial innovation project.

M2R / MSc internships

Quentin Guilloteau

« Minimizing Cluster Under-use using a Control-Based Approach »

Abstract:
High Performance Computing (HPC) systems have become increasingly more complex in the last decade. Their behaviour concerning their per- formance and power consumption make them less predictable. This unpre- dictability requires more and more runtime management. We believe that the field of Control Theory can bring a new approch to the administration of such complex systems. This project integrates an autonomic approach to the CiGri middleware, a grid application to manage large sets of multi-parametric tasks.
This work presents two distinct contributions. In a first time, we present a solution to minimize cluster under-utilization while controlling the load of the fileserver. We then investigate the case of reproducibility for our studied distributed system.

Manal BENAISSA

« Relationships between Scheduling and Autonomic Computing Techniques Applied to Parallel Computing Resource Management »

Abstract:
The scheduling field regroups various methods by which work is distributed across available computational resources. Considering the complexity of modern infrastructures, particularly in Cloud and HPC computing, schedulers might face difficulties to propose an efficient scheduling while fitting to the system reality and it implied uncertainties. The autonomic computing field suggests a more practical approach, by controlling constantly a system and adjusting taken decisions at runtime, via feedback loops. Applying this strategy in the scheduling context bring a less complex model with a more modular structure. Likewise, scheduling community may bring a new vision of autonomic computing challenges. This work presents some possible models, from the most basic scheduling algorithm, to the most complex Cloud infrastructures.

Gwenaël Delaval (UGA)

« Discrete Control of Response for Cybersecurity in Industrial Control »

slides

Abstract:
Cybersecurity in Industrial Control Systems (ICS) is a crucial problem, as recent history has shown. A notable characteristic of ICS, compared to Information Technology, is the necessity to take into account the physical process, and its specific dynamics and effects on the environment, when considering cybersecurity issues. Intrusion Detection Systems have been studied extensively. In our work, we address the less classic topic of response mechanisms, and their automation in a self-protection feedback loop. More precisely, we address self-protection seen as resilience, where the functionality of the system is maintained under attacks, be it in a degraded mode. We model this as a Discrete Event Systems supervisory control problem, involving a model of the plant’s possible behaviors, a model of considered attacks, and a formulation of the control objectives. We consider a case study, and perform a prototype implementation and simulation, using the Heptagon/BZR programming language and compiler/code generator, and targeting a multi-PLC experimental platform.

the Ctrl-A team

« one-minute madness presentations (ii) »

Second session of the one-minute madness presentations of each of the research topics in the team, by non-permanents as well as permanents. More precisely : each of us presents two slides max on her/his current ongoing topic (no results yet 😉 ) and I can make an overall presentation of the team to show how everything is related.

For this second session :

• Manal Benaissa : “Relationships between Scheduling Techniques in Parallel Computing and Feedback loops in Autonomic Computing” slides
• Raphaël Bleuse : ” Towards context-awareness for HPC resources management: models, software architectures, …” slides
• Sophie Cerf : “Combining Control Theory and Machine Learning in all ways: state of the art review and perspectives of application on Software-Defined Network” slides
• Stéphane Mocanu : “Cybersecurity of Industrial Control systems (SCADA)” slides
• Mahyar T. Moghaddam : “Self-adaptive middleware support for IoT/CPS” slides
• Lucie Muller : “Discrete controller synthesis: how can we know if it is useful ? (ii) ” slides (same as Gwenaël’s the week before)
• Eric Rutten : “Ctrl-A in 180 seconds” (again) slides (same as the week before)

the Ctrl-A team

« one-minute madness presentations »

the first session of the one-minute madness presentation of each of the research topics in the team, by non-permanents as well as permanents. More precisely : each of us presents two slides max on her/his current ongoing topic (no results yet 😉 ) and I can make a 2-slides overall presentation of the team to show how everything is related.

For this first session :

• Neil Ayeb : “Autonomic IoT Device Management” slides
• Sophie Cerf : “Control of RAPL: Energy-efficient HPC through processor-level power capping” slides
• Gwenaël Delaval : “Discrete controller synthesis: how can we know if it is useful ?” slides
• Quentin Guilloteau : “Minimizing Cluster Under-Use with a Control-Based Approach” slides
• Bashir Ibrahim : “Control-theory based Cluster Resources Optimization” slides
• Eric Rutten : “Ctrl-A in 180 seconds” slides

Sophie Cerf (post-doc, Inria)

« Why using Control theory approach for Computing systems. Example of resources allocation on the cloud »

slides

Abstract:
This talk is not a mathematical introduction to the control theory but rather an overview of its hypothesis, objectives, tools and limitations from a computer scientist point of view.
After motivating why you should include the control perspective in your researches, we’ll quickly review what is control theory, from its first applications during the industrial revolution, to the new advances that make our state of the art technologies.
In a few words, control theory deals with dynamical systems evolving through time, and aims at control them to bring them to a desired state, with given properties such as precision, rapidity and robustness. A control scientist is backed up by a strong theoretical framework with tools such as feedback loops, controllers, models, observers and frequency analysis.
A running example of a cloud’s resources allocation controller will be used to illustrate the introduced notions.

the previous lives of our new team members : short talks from M2Rs about their M1

• Manal Benaissa (Inria, UGA M2R MOSIG intern) « Development of tracing tools and trace analysis for OpenMP » Abstract:
  Simulation in research domains requires more and more complex and time-consuming algorithms. Many of these problems can be easily parallelized through all CPU to reduce execution time. OpenMP is a programming model specialized in shared-memory parallelism. One of its characteristics is the possibility to support tasks paradigm. In 2018, OpenMP released its 5th version, with an interesting API to implement a customized tracing tool: OMPT. The subject of this internship was to test the limit of this new feature in tasks context, by implementing suited tracing tool. Tests included results conversion in readable format and trace analysis.
• Quentin Guilloteau (Inria, Ensimag M2R MOSIG intern) « Adaptive Parallel Mergesort in Rust » Abstract:
  Sorting is one of the most fundamental operations in computer science. However, most of the sorting algorithms are sequential and do not use the multi-core processors of the modern machines. The mergesort is a classic example of “divide and conquer” algorithm and is embarrassingly parallel. Parallel algorithms should not depend on the architecture of the machines they run on. That is why techniques, such as adaptive algorithms, have been developed to make abstraction of the architecture of the machines. Adaptive algorithms are algorithms that can change their behavior based on various parameters: size of the input, number of threads in the system … This work consists of the first steps of implementing an adaptive mergesort algorithm in Rust. We implemented an adaptive version of the “classical” mergesort that only divides the work in two parts, and optimized it from the results of our experiments. We then compared its performance to the sorting algorithms of standard parallel libraries. We finally investigated the potential gain in performance by splitting the work in 3 parts instead of 2.

Mahyar Tourchi Moghaddam (Inria, H2020 CPS4EU)

« IoT-based Crowd Monitoring and Emergency Handling »

Abstract:
Cultural Heritage is a key element in fostering the socio-cultural transformation of citizens. Whilst the attractiveness of venues (such as museums) has positive impacts in cultural and economic areas, their management remains problematic. Crowd density is one of the most influential factors that negatively affects visitors’ experience. Such issue can be solved by designing human-focused Internet of Thing (IoT) systems for crowd management.

This talk deals with two aspects of crowd management in hyper-congested museums:

1. Analyzing human behavior in ticket booking, queuing and visiting the museum in order to design high-quality software/hardware architectures for sustainable queue management.

2. Monitoring visitors’ movement to design IoT-based emergency handling systems which provide real-time evacuation instructions.

Both systems take advantage of optimization algorithms as the core of their software architectures in order to both remove the long queues and (if needed) evacuate the occupants as quickly as possible.

The main topics to be discussed are: IoT software architectures, optimization algorithms and social agent-based modeling.

http://www.gipsa-lab.fr/summerschool/auto2019/

Self-managing or autonomic computing systems are answering to the need to address dynamic variations in the computing, memory or communication loads, as well as in their environment, the evolutions in their computing infrastructure (shared or subject to faults) or (re)adaptations of their initial functionalities. Their administration, usually performed by human administrators, needs to be automated in order to be efficient, safe and highly reactive. The Autonomic Computing paradigm using self-manageable closed loops emerged in the early 2000, targeting distributed system and addressing these questions from a computer science point of view. A particularly insightful way of building such control loops is to use control systems theory, which employs a large spectrum of modelling, estimation and control techniques (continuous, discrete, stochastic), classically applied mostly to electro-mechanical, physical systems, but much less usually to computing systems. .

The aim of this Summer School is to offer the opportunity of a scientific forum from control systems, informatics, distributed systems, around the various challenges and methodologies dedicated to the control of computing systems. To this end, domain experts will be present to share their expertise and cutting-edge research results.

• Ayan Hore (M2R MOSIG) ; advisors : S. Mocanu, E. Rutten
  « Using feedback loop to automate security response in Industrial Control Systems (ICS) »
• Abdul-Hafeez ALI (M2R MISCIT) ; advisors : O. Richard (LIG/Inria Datamove), B. Robu (Gipsa-lab), E. Rutten
  « Controlling Computer Cluster Overload, Intelligent Control With Non Supervised Learning »
• Mohsen Zargarani (M2R MISCIT) ; advisor : S. Mocanu
  « Hardware in the loop, Simulation of physical processes »

Lieu de soutenance de la thèse: Salle Chartreuse, GIPSA-lab – Domaine universitaire – 11 Rue des Mathématiques – 38400 Saint-Martin-d’Hères

Thèse dirigée par :
– THIRIET Jean-Marc, GIPSA-lab directeur de thèse,
– MOCANU Stéphane, LIG
– HIET Guillaume, CentraleSupélec

RÉSUME DE THÈSE

L’objectif de ce travail de thèse est le développement de techniques de
détection d’intrusions et de corrélation d’alertes spécifiques aux
systèmes de contrôle industriels (ICS). Cet intérêt est justifié par
l’émergence de menaces informatiques visant les ICS, et la nécessité de
détecter des attaques ciblées dont le but est de violer les
spécifications sur le comportement correct du processus physique. Dans
la première partie de nos travaux, nous nous sommes intéressés à
l’inférence automatique de spécifications pour l’aspect séquentiel des
ICS, et ce, à des fins de détection d’intrusions. Dans notre approche,
nous avons adopté le formalisme de la logique temporelle linéaire (LTL)
et métrique (MTL) permettant de représenter des propriétés temporelles
d’ordre qualitatif et quantitatif sur l’état des actionneurs et des
capteurs. Un algorithme d’inférence de propriétés a été développé afin
d’automatiser la génération des propriétés à partir de motifs de
spécifications couvrant les contraintes les plus communes. Cette
approche vise à pallier le nombre conséquent de propriétés redondantes
inférées par les approches actuelles. Dans la deuxième partie de nos
travaux, nous cherchons à combiner l’approche de détection d’intrusions
développée dans le premier axe avec des approches de détection
d’intrusions classiques. Pour ce faire, nous explorons une approche de
corrélation tenant compte des spécificités des systèmes industriels en
deux points : (i) l’enrichissement et le prétraitement d’alertes
hétérogènes venant de domaines différents (cyber et physique), et (ii)
la mise au point d’une politique de sélection d’alertes tenant compte du
contexte d’exécution du processus physique. Les résultats de
l’évaluation de nos approches de détection et de corrélation montrent
des performances améliorées sur la base de métriques telles que le taux
de réduction des alertes et la complétude des corrélations.

The objective of this thesis is to develop intrusion detection and alert
correlation techniques geared towards industrial control systems (ICS).
Our interest is driven by the recent surge in cybersecurity incidents
targeting ICS, and the necessity to detect targeted attacks which induce
incorrect behavior at the level of the physical process. In the first
part of this work, we develop an approach to automatically infer
specifications over the sequential behavior of ICS. In particular, we
rely on specification language formalisms such as linear temporal logic
(LTL) and metric temporal logic (MTL) to express temporal properties
over the state of the actuators and sensors. We develop an algorithm to
automatically infer specifications from a set of specification patterns
covering the most recurring properties. Our aim is to reduce the number
of redundant and unfalsifiable properties generated by the existing
approaches. In the second part of this work, we attempt to combine the
physical domain intrusion detection approach developed in the first part
with more classical cyber domain intrusion detection solutions. In
particular, we develop an alert correlation approach which takes into
account some specificities of ICS through: (i) the enrichment and
pre-treatment of heterogeneous alerts coming from different domains
(physical and cyber), and (ii) the specification of an alert policy
taking into account the execution context of the physical process. Our
evaluation results show that our approaches achieve improved results
with respect to classical detection and correlation metrics.

MEMBRES DU JURY

Jean Marc THIRIET – Directeur de these, Université Grenoble Alpes
Isabelle CHRISMENT – Rapporteur, Université de Lorraine
Michel COMBACAU – Rapporteur, Université Toulouse III Paul Sabatier
Isabel DEMONGODIN – Examinateur, Université d’Aix-Marseille
Gregory BLANC – Examinateur, Institut Mines-Télécom
Guillaume HIET – Examinateur, CentraleSupélec
Stéphane MOCANU – Examinateur, Grenoble INP
Frédéric MAJORCZYK – Examinateur, DGA

Enforcing a well-mastered and safe behaviour of the overall system, in presence of these concurrent adaptations, is a complex control problem. We approach this problem by applying techniques from the area of Supervisory Control for Discrete Event Systems (DES), where the space of configurations at the different levels (applications, tasks, execution platform) are modeled with automata. We use programming language support tools, Heptagon/BZR and ReaX, to build up a design environment for the considered application domain. We exploit especially modular and hierarchical control structures, in order to cope with the size of systems, and also logico-numerical control, in order to obtain controllers that can safely react not only to events but also to quantitative inputs.

This work contributes with (i) generic modeling of the behaviors and objectives for the considered systems; (ii) applications of Discrete Controller Synthesis (DCS) to design controllers, ; (iii) implemented case studies.

Such problem is found in the context of CiGri, a simple, lightweight, scalable and fault tolerant grid system which exploits the unused resources of a set of computing clusters. Computing power left over by the execution of a main HPC application scheduling is used to execute smaller jobs, which are injected as much as the global system allows.

The seminar will presents results addressing the problem of automated resource management in an HPC infrastructure, using techniques from Control Theory to design a controller that maximizes cluster utilization while considering cluster and fileserver overloading. We draw from a previous work where a proportional-integral feedback (Proportional Integral, PI) control loop was implemented, through a maximum number of jobs to be sent to the cluster, in response to system information about the current number of jobs processed. Then, we proposed a dynamical model for the system queue, cluster and fileserver load with time-varying parameters, an EKF (Extended Kalman Filter) for performing online parameter estimation and an MPC (Model-Predictive Control) approach for regulating the desired values of the infrastracture to the desired ones.

« Cybersecurity of SCADA and autonomic systems »

Abstract:
This talk will present the general setup of intrusion detection, protection and resilience of SCADA systems with en emphasis on process oriented intrusion detection and resilience.
The aim is to point out the possible links with self-protecting and self-reconfiguring systems as a reaction and resilience solution in case of attacks.

« Support intergiciel pour la conception et le déploiement adaptatifs fiables, application aux bâtiments intelligents »

Abstract:

Dans le contexte de l’informatique pervasive et de l’internet des objets, les systèmes sont hétérogènes, distribués et adaptatifs (p. ex., systèmes de gestion des transports, bâtiments intelligents). La conception et le déploiement de ces systèmes sont rendus difficiles par leur nature hétérogène et distribuée mais aussi le risque de décisions d’adaptation conflictuelles et d’inconsistances à l’exécution. Les inconsistances sont causées par des pannes matérielles ou des erreurs de communication. Elles surviennent lorsque des actions correspondant aux décisions d’adaptation sont supposées être effectuées alors qu’elles ne le sont pas.

Cette thèse propose un support intergiciel, appelé SICODAF, pour la conception et le déploiement de systèmes adaptatifs fiables. SICODAF combine une fiabilité comportementale (absence de décisions conflictuelles) au moyen de systèmes de transitions et une fiabilité d’exécution (absence d’inconsistances) à l’aide d’un intergiciel transactionnel. SICODAF est basé sur le calcul autonomique. Il permet de concevoir et de déployer un système adaptatif sous la forme d’une boucle autonomique qui est constituée d’une couche d’abstraction, d’un mécanisme d’exécution transactionnelle et d’un contrôleur. SICODAF supporte trois types de contrôleurs (basés sur des règles, sur la théorie du contrôle continu ou discret). Il permet également la reconfiguration d’une boucle, afin de gérer les changements d’objectifs qui surviennent dans le système considéré, et l’intégration d’un système de détection de pannes matérielles. Enfin, SICODAF permet la conception de boucles multiples pour des systèmes qui sont constitués de nombreuses entités ou qui requièrent des contrôleurs de types différents. Ces boucles peuvent être combinées en parallèle, coordonnées ou hiérarchiques.

SICODAF a été mis en œuvre à l’aide de l’intergiciel transactionnel LINC, de l’environnement d’abstraction PUTUTU et du langage Heptagon/BZR qui est basé sur des systèmes de transitions. SICODAF a été également évalué à l’aide de deux études de cas.

« Détection d’intrusions et corrélation d’alertes dans les systèmes de contrôle séquentiels »

Abstract:
Du fait des menaces de cybersécurité auxquelles font face les systèmes de contrôle industriels (ICS), ainsi que de l’insuffisance des mécanismes de protection, le développement d’approches de détection d’intrusions adaptées devient crucial. En particulier, le processus physique constitue une cible privilégiée pour les attaquants. Cependant, les approches actuelles de détection d’intrusions ne permettent pas une contextualisation des alertes par rapport à l’état du processus physique. Par conséquent, les opérateurs de sécurité peinent à distinguer entre attaques et faux positifs dus à des interventions manuelles légitimes. Nos travaux cherchent à répondre à ce besoin en développant des approches de détection d’intrusions et de corrélation d’alertes tenant compte des aspects cyber et physique des ICS, notamment dans le cadre peu exploré des attaques visant les systèmes de contrôle séquentiels.

« Techniques de contrôle pour la gestion autonomique des plateformes de calcul »

Abstract:
Cet exposé présente un travail sur les techniques de contrôle pour la gestion autonomique des plateformes de calculs. On considère en particulier le mésocentre de calcul intensif de l’Université Grenoble-Alpes : le mésocentre “Ciment”. L’objectif consiste à mettre en œuvre une régulation afin de maximiser l’utilisation des ressources sans entraîner de surcharge. Après analyse du fonctionnement du système, différents tests en boucle ouverte afin d’observer son comportement. Ceci permet d’identifier un modèle, qui est utilisé dans la mise en place d’une correction Proportionnelle – Intégrale. Des tests expérimentaux sont effectués pour valider cette régulation.

« A Language for the Smart Home »

Abstract:
This work aims to propose a end user development language which allows inhabitants of a smart home to express the behaviours they want to their home automation system. This language should bring together the best features of existing ECA-based solutions and missing properties that are asked by current users or highlighted by research works.

« Application-aware control for dynamically reconfigurable architectures »

Abstract:
Dynamic Partial Reconfiguration (DPR) available in modern FPGA allows an FPGA device to support more hardware available than physically possible. It enables dynamic reconfiguration of part of an FPGA at run-time without having to reconfigure the entire device. The unmodified regions can continue executing their computation tasks without any interruption during partial reconfiguration.

DPR is a promising solution for embedded systems that require high performance and high flexibility while achieving energy efficiency and minimal footprints. This is typically the case of small autonomous UAVs. These systems mostly require to be self-adaptive to address uncertainties related to the environment. Self-adaptation can require to replace the computation tasks at runtime according to observations. Hence, there is need for a decision making component that manages at runtime the tasks to execute and the hardware reconfiguration.

In this presentation we will describe ongoing work on our approach for designing a control architecture for the management of the reconfiguration of the hardware architecture at run-time.

« Autonomic Thread Parallelism and Mapping Control for Software Transactional Memory »

Abstract:
Parallel programs need to manage the trade-off between the time spent in synchronisation and computation. The trade-off is significantly affected by the number of active threads. High parallelism may decrease computing time while increase synchronisation cost. Furthermore, thread placement on different cores may impact on program performance, as the data access time can vary from one core to another due to intricacies of its underlying memory architecture. Therefore, the performance of a program can be improved by adjusting its parallelism degree and the mapping of its threads to physical cores. Alas, there is no universal rule to decide them for a program from an offline view, especially for a program with online behaviour variation. Moreover, offline tuning is less precise. This thesis presents work on dynamical management of parallelism and thread placement. It addresses multithread issues via Software Transactional Memory (STM). STM has emerged as a promising technique, which bypasses locks, to tackle synchronisation through transactions. Autonomic computing offers designers a framework of methods and techniques to build autonomic systems with well-mastered behaviours. Its key idea is to implement feedback control loops to design safe, efficient and predictable controllers, which enable monitoring and adjusting controlled systems dynamically while keeping overhead low. This dissertation proposes feedback control loops to automate management of threads at runtime and diminish program execution time.

Jury :

M.Raymond Namyst: Professeur, Université de Bordeaux, Rapporteur
M.Lionel Seinturier: Professeur, Université Lille 1, Rapporteur
M.Christian Perez: Directeur de recherche, INRIA Lyon , Examinateur
M.Jean-François Méhaut: Professeur, Université de Grenoble Alpes, Directeur de thèse
M.Éric Rutten: Chargé de recherche, INRIA Grenoble , Directeur de thèse
M.Gwenaël Delaval: Associate professeur, Université de Grenoble Alpes, Directeur de thèse

« Smart home applications control and reactive programming
A service-oriented approach »

Abstract :

Avec un nombre sans cesse croissant de moyens de communications divers, le besoin d’adaptabilité en informatique ubiquitaire se fait d’autant plus sentir. Pour des applications autonomes, les architectures de contrôle deviennent alors elle-même très complexes, requérant également des adaptations

Les applications autonomes sont souvent composées de plusieurs boucles de contrôle – chacune adressant un besoin spécifique – dont l’exécution nécessite d’être coordonnée pour une administration efficace et correcte. Nous proposons donc d’étudier l’usage de modèles de contrôle réactifs avec des événements et états pour coordonner des boucles autonomiques dans des architectures orientées services.

Dans cet ouvrage, nous illustrons notre approche en intégrant un contrôleur basé sur la synthèse de contrôleurs discrets dans un environnement ubiquitaire autonome. Le rôle du contrôleur est d’influencer les critères de liaison des services de multiples boucles de contrôle, tout en respectant des contraintes logiques. Nous considérons particulièrement les opérations de reconfiguration d’ensembles des services connus et dynamiques.

Nos résultats suggèrent que, même si le temps de synthèse et la taille finale du contrôleur augmentent rapidement avec le nombre des états et entrées, les systèmes réactifs sont une approche appropriée de coordination pour les applications orientées services.

« Cybersecurity of Industrial Control Systems »

Abstract :

We present some ongoing research on the cybersecurity of SCADA systems. We present the specificity and challenges of the Industrial Systems cybersecurity then we focus on Intrusion Detection Systems problem and present our approach. In the last part we’ll try to point out the links between detection and reaction in SCADA systems and autonomic systems and supervisory control.

« Combining Reactive Coordination Rules and Automata Languages for Safe Smart Environments »

Abstract :

This work combines two reactive programming approaches
in order to support the safe design and execution of control systems,
in the application domain of smart environments. Rule-based languages,
like the one in the coordination environment LINC, provide for high-level
programming of smart environments behaviours. LINC also ensures safe
execution supported by transactional management of rules. However,
when the rules are numerous, they can be inconsistent or bring the system
into unsafe states. On the other hand, reactive languages based on
automata, enable verification, and even synthesis of controllers in the case
of Heptagon/BZR. This paper studies these two languages and combines
their reactive execution mechanisms. A case study taken in the field of
building automation is presented to illustrate the proposed approach.

« Autonomic Parallelism and Thread Mapping Control on Software Transactional Memory »

Inria Montbonnot, C208

Abstract :

Parallel programs need to manage the time trade-off between synchronization and computation. The time trade-off is hugely affected by the number of active threads. A high parallelism may decrease computing time meanwhile increase synchronization cost. Additionally threads locality on different cores may impact on program performance too, as the memory access time can vary from one core to another due to the complexity of the underlying memory architecture. Therefore a program performance can be improved by adjusting the number of active threads as well as their localities. However, there is no universal rule to decide the parallelism and the thread locality for a program from an offline view. Furthermore, an offline tuning is costly and error-prone. In this paper, we dynamically adjust parallelism and thread localities. We address multiple threads issues via Software Transactional Memory (STM). STM has emerged as a promising technique, which bypasses locks, to address synchronization issues through transactions. Autonomic computing offers designers a framework of methods and techniques to build autonomic systems with well-mastered behaviours. Its key idea is to implement feedback control loops to design safe, efficient and predictable controllers, which enable monitoring and adjusting controlled systems dynamically while keeping overhead low. We propose to design a feedback control loop to automate thread management at runtime and diminish program execution time.

« Control and Performance Evaluation of Computing Systems in Presence of Uncertainty »

DAUTO – salle des Séminaires B208, Gipsa-lab, Campus
http://www.gipsa-lab.grenoble-inp.fr/plan-d-acces.php

Abstract :

The increasing complexity in different areas of computing systems calls for novel inter-disciplinary approaches, able to cope with dynamic and uncertain environments. Different decision-making mechanisms are considered in the literature, but the robustness of the proposed techniques is proven only on a limited number of cases, without providing general guarantees. The aim of this talk is to present: 1) an overview of simple control techniques that can be applied to relevant problems in computing systems, 2) a performance assessment methodology that gives probabilistic guarantees of the propose techniques in presence of uncertainty. These concepts will be explained via a running example on load-balancing techniques for cloud computing.

« Autonomic Parallelism Adaptation on Software Transactional Memory »

DAUTO – salle des Séminaires B208, Gipsa-lab, Campus
http://www.gipsa-lab.grenoble-inp.fr/plan-d-acces.php

« A Reconfigurable Component Model for HPC »

DAUTO – salle des Séminaires B208, Gipsa-lab, Campus
http://www.gipsa-lab.grenoble-inp.fr/plan-d-acces.php

« Contrôle Discret pour l’Internet des Objets et les Environnements Intelligents au travers d’une infrastructure partagée »

Orange Labs, 28 chemin du vieux chêne, 38240 Meylan

PhD defense (in french)

« Coordination Rules Generation from Coloured Petri Net Models »

DAUTO – salle des Séminaires B208, Gipsa-lab, Campus
http://www.gipsa-lab.grenoble-inp.fr/plan-d-acces.php

« Future technologies and the challenge of high speed, low energy, high reliability computing: what are the challenges? »

DAUTO – salle des Séminaires B208, Gipsa-lab, Campus
http://www.gipsa-lab.grenoble-inp.fr/plan-d-acces.php

Soguy Mak-Karé Gueye

« Coordination Modulaire de Gestionnaires Autonomes par Contrôle Discret »

PhD defense (in french)

« Designing Autonomic Management Systems by using Reactive Control Techniques »
work with : Soguy Mak-Karé Gueye, Noël de Palma, Eric Rutten

sams

modular-coordination

work with : Bogdan Robu (Gipsa-lab), Sara Bouchenak (LIG), Damian Serano (LIG INRIA) et Nicolas Marchand (Gipsa-lab)

Mengxuan Zhao (Orange labs)

« Modèles génériques applicables à la synthèse de contrôleurs discrets pour l’internet des objets »

« Self-healing of workflow executions: application to late task
execution, task granularity, and fairness among workflows »

« (Towards a) Reactive Design of Autonomic Management Systems »

Towards distributed monitoring (of Android applications)

Mengxuan Zao

premières approches en contrôle supervisé dans le “smart home”

• Soguy Gueye
  Coordinating energy-aware administration loops
  using discrete control
• Xin An
  Discrete Controller Synthesis for Runtime
  Management of Dynamically Reconfigurable
  Embedded Systems