CTRL-A: Commande pour systèmes informatiques autonomiques

With the growing availability of large-scale datasets, and the popularization of affordable storage and computational capabilities, the energy consumed by AI is becoming a growing concern. To address this issue, in recent years, studies have focused on demonstrating how AI energy efficiency can be improved by tuning the model training strategy. Nevertheless, how modifications applied to datasets can impact the energy consumption of AI is still an open question.To fill this gap, in this exploratory study, we evaluate if data-centric approaches can be utilized to improve AI energy efficiency. To achieve our goal, we conduct an empirical experiment, executed by considering 6 different AI algorithms, a dataset comprising 5,574 data points, and two dataset modifications (number of data points and number of features).Our results show evidence that, by exclusively conducting modifications on datasets, energy consumption can be drastically reduced (up to 92.16%), often at the cost of a negligible or even absent accuracy decline. As additional introductory results, we demonstrate how, by exclusively changing the algorithm used, energy savings up to two orders of magnitude can be achieved.In conclusion, this exploratory investigation empirically demonstrates the importance of applying data-centric techniques to improve AI energy efficiency. Our results call for a research agenda that focuses on data-centric techniques, to further enable and democratize Green AI.

This paper presents a scalable approach to model uncertainties within a UAV (Unmanned Aerial Vehicle) embedded mission manager. It proposes a concurrent version of BFM models, which are Bayesian Networks built from FMEA (Failure Mode and Effects Analysis) and used by MDPs (Markov Decision Processes). The models can separately handle different applications during the mission; they consider the context of the mission including external constraints (luminosity, climate, etc.), the health of the UAV (Energy, Sensor) as well as the computing resource availability including CPU (Central Processing Unit) load, FPGA (Field Programmable Gate Array) use and timing performances. The proposed solution integrates the constraints into a mission specification by means of FMEA tables in order to facilitate their specifications by non-experts. Decision-making processes are elaborated following a "just enough" quality management by automatically providing adequate implementation of the embedded applications in order to achieve the mission goals, in the context given by the sensors and the on-board monitors. We illustrate the concurrent BFM approach with a case study of a typical tracking UAV mission. This case also considers a FPGA-SoC (FPGA-System on Chip) platform into consideration and demonstrates the benefits to tune the quality of the embedded applications according to the environmental context.

The increasing natural and man-induced disasters such as res, earthquakes, oods, hurricanes, overcrowding, or pandemic viruses endanger human lives. Hence, designing infrastructures to handle those possible crises has become an ever-increasing need. The Internet of Things (IoT) has changed our approach to safety systems by connecting sensors and providing real-time data to managers, rescuers, and endangered people. IoT systems can monitor and react to progressive disasters, people's movements and their behavioral patterns. The community faces challenges in using IoT for crises management: i) how to take advantage of technological advancements and deal with IoT resources installation issues? ii) what environmental contexts should be considered while designing IoT-based emergency handling systems? iii) how should system design comply with various levels of real-time requirements? This paper reports on the results of the First International Workshop on Internet of Things for Emergency Management (IoT4Emergency 2020), which speci cally focuses on challenges and envisioned solutions in using smart connected systems to handle disasters.

Modern power-network communications are based on the IEC 61850 series standards. In this paper, we investigate the real-time performance and the vulnerabilities and attack scenarios at the sensor level communication networks more precisely on Sampled Measured Value protocol. The approach jointly evaluates the communication protocol, network topology and impact on electrical protection functions. We test the practical feasibility of the attacks on an experimental workbench using real devices in a hardware-in-the-loop setup. The tests are conducted on the two high-availability automation networks currently used in IEC 61850 process bus communications: Parallel Redundancy Protocol (PRP) and High-availability Seamless Redundancy (HSR)

Dynamically reconfigurable hardware has been identified as a promising solution for the design of energy-efficient embedded systems. However, its adoption is limited by costly design effort, including verification and validation, which is even more complex than for nondynamically reconfigurable systems. In this article, we propose a tool-supported formal method to automatically design a correct-by-construction control of the reconfiguration. By representing system behaviors with automata, we exploit automated algorithms to synthesize controllers that safely enforce reconfiguration strategies formulated as properties to be satisfied by control. We design generic modeling patterns for a class of reconfigurable architectures, taking into account both hardware architecture and applications, as well as relevant control objectives. We validate our approach on two case studies implemented on FPGAs.

This article presents the first framework to design and synthesize a formal controller managing dynamic reconfiguration, using a model-driven engineering methodology based on an extension of UML/MARTE. The implementation technique highlights the combination of hard configuration constraints using weights ( control part )—ensured statically and fulfilled by the system at runtime—and soft constraints ( decision part ) that, given a set of correct and accessible configurations, choose one of them. An application model of an image processing application is presented, then transformed and synthesized to be executed on a Xilinx platform to show how the controller, executed on a Microblaze, manages the hardware reconfigurations.

Industrial Control Systems (ICS) are specific systems that combine information technology (IT) and operational technology (OT). Due to their interconnection and remote accessibility, they become a target for cyberattacks. As a result of their complexity and heterogeneity in terms of devices and communication protocols, specific security controls and risk analysis methods need to be developed. In particular, in order to reduce the effort of deployment of risk analysis on such complex systems, automated methods need to be provided. This paper deals with automation of the risk identification process for ICS using the STRIDE threat modeling framework. We extend the well-known STRIDE modeling tool, namely Microsoft Threat Modeling Tool (MTMT), with an incremental template dedicated to ICS and provide additional tools to automate the analysis using specific vulnerability extraction from Internet CVE databases.

The ever growing complexity of software systems has led to the emergence of automated solutions for their management. The software assigned to this work is usually called an Autonomic Management System (AMS). It is ordinarily designed as a composition of several managers, which are pieces of software evaluating the dynamics of the system under management through measurements (e.g., workload, memory usage), taking decisions, and acting upon it so that it stays in a set of acceptable operating states. However, careless combination of managers may lead to inconsistencies in the taken decisions, and classical approaches dealing with these coordination problems often rely on intricate and ad hoc solutions. To tackle this problem, we take a global view and underscore that AMSs are intrinsically reactive, as they react to flows of monitoring data by emitting flows of reconfiguration actions. Therefore we propose a new approach for the design of AMSs, based on synchronous programming and discrete controller synthesis techniques. They provide us with high-level languages for modeling the system to manage, as well as means for statically guaranteeing the absence of logical coordination problems. Hence, they suit our main contribution, which is to obtain guarantees at design time about the absence of logical inconsistencies in the taken decisions. We detail our approach, illustrate it by designing an AMS for a realistic multi-tier application, and evaluate its practicality with an implementation.

In this paper, we present a SCADA cybersecurity awareness and training program based on a Hands-On training using two twin cyber-ranges named WonderICS and G-ICS. These labs are built using a Hardware-In-the-Loop simulation system of the physical process developed by the two partners. The cyber-ranges allow replication of realistic Advanced Persistent Threat (APT) attacks and demonstration of known vulnerabilities, as they rely on real industrial control devices and softwares. In this work, we present both the demonstration scenarios used for awareness on WonderICS and the training programs developed for graduate students on G-ICS.

Abstract Correctness of the behavior of an adaptive system during dynamic adaptation is an important challenge to realize correct adaptive systems. Dynamic adaptation refers to changes to both the functionality of the computational entities that comprise a composite system, as well as the structure of their interconnections, in response to variations in the environment, e.g., the load of requests on a server system. In this research, we view the problem of correct structural adaptation as a supervisory control problem and synthesize a reconfiguration controller that guides the behavior of a system during adaptation. The reconfiguration controller observes the system behavior during an adaptation and controls the system behavior by allowing/disallowing actions in a way to ensure that a given property is satisfied and a deadlock is avoided. The system during adaptation is modeled using a graph transition system and properties to be enforced are specified using a graph automaton. We adapt a classical theory of supervisory control for synthesizing a controller for controlling the behavior of a system modeled using graph transition systems. This theory is used to synthesize a controller that can impose both behavioral and structural constraints on the system during an adaptation. We apply a tool that we have implemented to support our approach on a case study involving https servers.

Cloud and HPC (High-Performance Computing) systems have increasingly become more varying in their behavior, in particular in aspects such as performance and power consumption, and the fact that they are becoming less predictable demands more runtime management. In this work, we describe results addressing autonomic administration in HPC systems for scientific workflows management through a control theoretical approach. We propose a model described by parameters related to the key aspects of the infrastructure thus achieving a deterministic dynamical representation that covers the diverse and time-varying behaviors of the real computing system. Later, we propose a model-predictive control loop to achieve two different objectives: maximize cluster utilization by best-effort jobs and control the file server's load in the presence of external disturbances. The accuracy of the prediction relies on a parameter estimation scheme based on the EKF (Extended Kalman Filter) to adjust the predictive-model to the real system, making the approach adaptive to parametric variations in the infrastructure. The closed loop strategy shows performance improvement and consequently a reduction in the total computation time. The problem is addressed in a general way, to allow the implementation on similar HPC platforms, as well as scalability to different infrastructures.

Dynamic reconfiguration is a key capability of Component-based Software Systems to achieve self-adaptation as it provides means to cope with environment changes at runtime. The space of configurations is defined by the possible assemblies of components, and navigating this space while achieving goals and maintaining structural properties is managed in an autonomic loop. The natural architectural structure of component-based systems calls for hierarchy and modularity in the design and implementation of composites and their managers, and requires support for coordinated multiple autonomic loops. In this paper, we leverage the modularity capability to strengthen the Domain-Specific Language (DSL) Ctrl-F, targeted at the design of autonomic managers in component-based systems. Its original definition involved discrete control-theoretical management of reconfigurations, providing assurances on the automated behaviors. The objective of modularity is two-fold: from the design perspective, it allows designers to seamlessly decompose a complex system into smaller pieces of reusable architectural elements and adaptive behaviours. From the compilation point of view, we provide a systematical and generative approach to decompose control problems described in the architectural level while relying on mechanisms of modular Discrete Control Synthesis (DCS), which allows us to cope with the combinatorial complexity that is inherent to DCS problems. We show the applicability of our approach by applying it to the self-adaptive case study of the existing RUBiS/Brownout eBay-like web auction system.

Industrial Control System cybersecurity has become an important study area after the occurrence of several mediatic events in the 2010’s (Stuxnet, BlackEnergy, Industroyer). Two common characteristics of these attacks are the fact that they were not violating the communication protocols being "stealth" for classical pattern-based detection methods and that they explicitly target the physical process. In this paper we study the performance and explainability of an artificial intelligence based detection system for the detection of such sophisticated attacks.

Parallel programs need to manage the trade-off between the time spent in synchronization and computation. The time trade-off is affected by the number of active threads significantly. High parallelism may decrease computing time while increase synchronization cost. Furthermore thread 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 the performance of a program can be improved by adjusting the number of active threads as well as the mapping of its threads to physical cores. 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 error-prone. In this paper, we dynamically manage parallelism and thread localities. We address multiple threads problems 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.

A smart environment is equipped with numerous devices (i.e., sensors, actuators) that are possibly distributed over different locations (e.g., rooms of a smart building). These devices are automatically controlled to achieve different objectives related, for instance, to comfort, security and energy savings. Controlling smart environment devices is not an easy task. This is due to: the heterogeneity of devices, the inconsistencies that can result from communication errors or devices failure, and the conflicting decisions including those caused by environment dependencies. This paper proposes a design framework for the reliable and environment aware management of smart environment devices. The framework is based on the combination of the rule based middleware LINC and the automata based language Heptagon/BZR (H/BZR). It consists of: an abstraction layer for the heterogeneity of devices, a transactional execution mechanism to avoid inconsistencies and a controller that, based on a generic model of the environment, makes appropriate decisions and avoids conflicts. A case study with concrete devices, in the field of building automation, is presented to illustrate the framework.

With the expansion of Internet of Things (IoT) that relies on heterogeneous, dynamic, and massively deployed devices, device management (DM) (i.e., remote administration such as firmware update, configuration, troubleshooting and tracking) is required for proper quality of service and user experience, deployment of new functions, bug corrections and security patches distribution.

In recent years, Informational Technologies (IT) was massively deployed into Industrial Control Systems (ICS) mainly for its economic benefits. However, this new paradigm, converging IT and Operational Technologies (OT), brings new challenges that companies need to face. Historically, ICS had to cope with safety requirements which ensure the protection of people, environment, and assets. Now, ICS must deal with additional threats, coming from cyberattacks, in order to maintain safety. For that purpose, it becomes essential to develop new cybersecurity technologies and methodologies that allow to assess the safety of ICS against cyberattacks.

Device Management (DM) is currently industrially deployed for LAN devices, phones and workstation management. Internet of Things (IoT) devices are massive, dynamic, heterogeneous, and inter-operable. Existing solutions are not suitable for IoT management. This doctoral research in an industrial environment addresses these limitations with a novel autonomic and distributed approach for the DM.

Modern power-network communications are based on the IEC 61850 series standards. In this paper we investigate the real-time performance, the vulnerabilities and the attack scenarios on the sensor level communication networks, more precisely on the Sampled Measured Value (SMV) protocol. There are two main contributions of our work. First, we evaluate statistically the measured real-time performance of the communication network. The second contribution is the description, implementation and experimental validation of the attacks on SMV protocol targeting electrical protection functions.

The feedback control of High-Performance Computing (HPC) has been explored as an application area of Control Theory, because of the high variability involved in their resource management. A regulation mechanism can allow to soundly automate the injection of small flexible jobs in a cluster. A trade-off is needed, to fill up the cluster’s computing capacity while avoiding overload of e.g., the file server.In this work, we describe new results in this context, where the overload avoidance controller is made adaptive to the jobs’ size, that is a time-varying unknown parameter. To do so, the original PI controller is enhanced with an online estimation algorithm that allows the controller to adapt to various working conditions, to avoid performance degradation. Parallel and robust estimation algorithms are designed, tackling the challenges of bursting and noise in the system. Validation and evaluation of the adaptive controller are performed on a large-scale experimental HPC platform, showing higher robustness than the state-of-the-art in highly varying conditions. Reproducible analysis are available at doi:10.5281/zenodo.11961696.