Search

Soumission à Episciences, Mobile ad-hoc social networks (MASNs) have been the subject of several research studies over the past two decades. They allow stations located in a small geographical area to be connected without the need for a network infrastructure and offer them the possibility to communicate any time anywhere. To communicate, stations regularly broadcast their interests in the form of keywords. Stations with a high degree of similarity among their keywords can communicate with each other. However, the coverage of MASNs is limited to a small geographical area, due to the limited communication range of mobile ad-hoc networks (MANET) stations. In this paper, we present an architecture and implementation of hybrid mobile ad-hoc social networks (MASNs coupled to infrastructure networks) of Android mobile devices for information dissemination. Stations can use the infrastructure network to communicate and rely on the mobile ad-hoc network when the infrastructure is not available.Rather than communicating synchronously as this is the case in the similar works found in the literature, in our approach, the stations communicate using a publish/subscribe communication protocol, which is perfectly suited to this type of network thanks to the decoupling in time and space it provides., Les réseaux sociaux mobiles ad hoc (MASN) ont fait l'objet de plusieurs recherches au cours des deux dernières décennies. Ils permettent à des stations situées dans une petite zone géographique d'être connectées sans avoir besoin d'une infrastructure de réseau et leur offrent la possibilité de communiquer à tout moment et en tout lieu. Pour communiquer, les stations diffusent régulièrement leurs centres d'intérêt sous forme de mots-clés. Les stations dont les mots-clés présentent un degré élevé de similitude peuvent communiquer entre elles. Cependant, la couverture des réseaux MASN est limitée à une petite zone géographique, en raison de la portée de communication limitée des stations des réseaux mobiles ad hoc (MANET). Dans cet article, nous présentons une architecture et une mise en œuvre de réseaux sociaux mobiles ad hoc hybrides (MASN couplés à des réseaux d'infrastructure) d'appareils mobiles Android pour la diffusion d'informations. Les stations peuvent utiliser le réseau d'infrastructure pour communiquer et s'appuyer sur le réseau mobile ad hoc lorsque l'infrastructure n'est pas disponible.Plutôt que de communiquer de manière synchrone comme c'est le cas dans les travaux similaires trouvés dans la littérature, dans notre approche, les stations communiquent en utilisant un protocole de communication de type publication/abonnement, qui est parfaitement adapté à ce type de réseau grâce au découplage dans le temps et l'espace qu'il fournit.

The spoligotype is a graphical description of the CRISPR locus present in Mycobacterium tuberculosis, which has the particularity of having only 68 possible spacers. This spoligotype, which can be easily obtained either in vitro or in silico, allows to have a summary information of lineage or even antibiotic resistance (when known to be associated to a particular cluster) at a lower cost. The objective of this article is to show that this representation is richer than it seems, and that it is under-exploited until now. We first recall an original way to represent these spoligotypes as points in the plane, allowing to highlight possible sub-lineages, particularities in the animal strains, etc. This graphical representation shows clusters and a skeleton in the form of a graph, which led us to see these spoligotypes as vertices of an unconnected directed graph. In this paper, we therefore propose to exploit in detail the description of the variety of spoligotypes using a graph, and we show to what extent such a description can be informative.

International audience; Failures in pipeline transportation of crude oil have numerous adverse effects, such as ecological degradation, environmental pollution and a decrease in revenue for the operators, to mention a few. Efficient data and service management can predict and prevent these failures, reducing the downtime of the pipeline infrastructure, among other benefits. Thus, we propose a two-stage approach to data and service management in Leakage Detection and Monitoring Systems (LDMS) for crude oil pipelines. It aims to maximise the accuracy of leakage detec- tion and localisation in a fault-tolerant and energy-efficient manner. The problem is modelled as a Markov Decision Process (MDP) based on the historical incident data from the Nigerian National Petroleum Corporation (NNPC) pipeline networks. Results obtained guarantee detection in at least two deployed nodes with a minimum localisation accuracy of 90%. Additionally, we achieved approximately 77% and 26% reduction in energy consumption compared to a pessimistic strategy and a globalised heuristic approach, respectively.

International audience; Self-adaptation of complex systems is a very active domain of research with numerous application domains. Component systems are designed as sets of components that may reconfigure themselves according to adaptation policies, which describe needs for reconfiguration. In this context, an adaptation policy is designed as a set of rules that indicate, for a given set of configurations, which reconfiguration operations can be triggered, with fuzzy values representing their utility. The adaptation policy has to be faithfully implemented by the system, especially w.r.t. the utility occurring in the rules, which are generally specified for optimizing some extra-functional properties (e.g. minimizing resource consumption). In order to validate adaptive systems’ behaviour, this paper presents a model-based testing approach, which aims to generate large test suites in order to measure the occurrences of reconfigurations and compare them to their utility values specified in the adaptation rules. This process is based on a usage model of the system used to stimulate the system and provoke reconfigurations. As the system may reconfigure dynamically, this online test generator observes the system responses and evolution in order to decide the next appropriate test step to perform. As a result, the relative frequencies of the reconfigurations can be measured in order to determine whether the adaptation policy is faithfully implemented. To illustrate the approach the paper reports on experiments on the case study of platoons of autonomous vehicles.

In this paper, we propose to address the multi-objective task scheduling issue in cloud computing system in order to find the best assignment of the user-submitted tasks to the cloud resources provided by the cloud service providers. To achieve a high overall performance for computational cloud resources, a good scheduler algorithm is required to meet customer requirements and ensure efficient QoS. To this end, the Flower Pollination Algorithm metaheuristic has been used to orchestrate the trade-offs relationship between three end-user QoS-based criteria, namely reliability, time makespan and execution cost. The results obtained from the experiment conducted with the CloudSim simulator clearly show the interest of our approach based on FPA.

Sensor nodes are one of the basic elements in the Tactile Internetbased fog computing architecture. They provide a huge amount of data to the network due to the widespread real-world applications that use these types of wireless devices. This huge number of data, transmitted by the sensor devices to the fog gateway and then to the cloud, leads to high communication costs, increased power consumption, and high latency at the fog gateway. These challenges would be considered as a hurdle in the Tactile Internetbased fog system. To tackle these challenges, this paper proposes an Energy-efficient Fo g Co mputing-enabled Da ta Transmission (EFoCoD) protocol in Tactile Internet-based Applications. The protocol works on sensor devices level in the Tactile Internet-based fog computing architecture. EFoCoD protocol executes a Lightweight Data Redundancy Elimination (LiDaRE) Algorithm at the sensor level to reduce the collected data before transferring them to the smart fog gateway. To study the performance of the EFoCoD, it was compared to its counterpart protocols in the literature just like ATP and PFF. Simulation results show that EFoCoD outperforms these protocols in terms of energy consumption, transmitted data, and data accuracy.

This study proposes machine learning-based models to automatically evaluate the severity of myocardial infarction (MI) from physiological, clinical, and paraclinical features. Two types of machine learning models are investigated for the MI assessment: the classification models classify the presence of the infarct and the persistent microvascular obstruction (PMO), and the regression models quantify the Percentage of Infarcted Myocardium (PIM) of patients suspected of having an acute MI during their reception in the emergency department. The ground truth labels for these supervised models are derived from the corresponding Delayed Enhancement MRI (DE-MRI) exams and manual annotations of the myocardium and scar tissues. Experiments were conducted on 150 cases and evaluated with cross-validation. Results showed that for the MI (PMO inclusive) and the PMO (infarct exclusive), the best models obtained respectively a mean error of 0.056 and 0.012 for the quantification, and 88.67 and 77.33% for the classification accuracy of the state of the myocardium. The study of the features' importance also revealed that the troponin value had the strongest correlation to the severity of the MI among the 12 selected features. For the proposal's translational perspective, in cardiac emergencies, qualitative and quantitative analysis can be obtained prior to the achievement of MRI by relying only on conventional tests and patient features, thus, providing an objective reference for further treatment by physicians.

Modeling Cyber-Physical Systems (CPS) remains a challenge due to the complex behaviors of their interconnected components that operate in a physical environment. In this paper, we introduce a new modeling approach based on "SysReo" models, that relies on SysML/MARTE, Reo, and OCL to capture the different aspects of CPS with time constraints, including requirements, behavior, architecture, and interaction protocols. The novelty of our approach relies in the combination of SysML and Reo to handle the complexity of CPS architecture and protocols, in the design step by proceeding incrementally. Then we used SysML/MARTE to model the behavior of CPS with timed constraints. Furthermore, we define OCL constraints to specify rules to be respected to model consistently CPS

Solid-State Nanopores made of 2-D materials such as MoS2 have emerged as one of the most versatile sensors for single-biomolecule detection, which is essential for early disease diagnosis (biomarker detection). One of the most promising applications of SSN is DNA and protein sequencing, at a low cost and faster than the current standard methods. The detection principle relies on measuring the relatively small variations of ionic current as charged biomolecules immersed in an electrolyte traverse the nanopore, in response to an external voltage applied across the membrane. The passage of a biomolecule through the pore yields information about its structure and chemical properties, as demonstrated experimentally particularly for DNA molecules. Indeed, protein sequencing using SSN remains highly challenging since the protein ensemble is far more complex than the DNA ensemble [1]. In this work, we performed extensive unbiased all-atom classical Molecular Dynamics simulations to produce data of translocation of biological peptides through single-layer MoS2 nanopores (D = 1.3 nm). Peptide made of 12 different amino acids from the different families (non polar/hydrophobic, polar/neutral, basic and acidic) were chemically linked to a short polycationic charge carrier. First, ionic current time series were computed from MD and peptide-induced blockade events were extracted and characterized using structural break detection. Second, clustering (unsupervised learning) of ionic current subdrops and

The transformation of cities into "smart cities" is one of the goals that governments have been aiming at for the last ten years. This transformation affects several aspects of the city, including eco-citizenship, sustainable governance, and smart mobility. It is on this aspect that our work focuses. Indeed, despite a certain progress in the last few years, people with reduced mobility are often forgotten when it comes to improving urban travel thanks to information and communication technologies. Among these technologies, artificial intelligence is booming and encompasses several fields. Among them, ambient intelligence aims at the design of interconnected services interacting with the real physical environment to provide services to support human activities. It is in this context that our work is placed: the study and construction of an ambient system dedicated to the assistance of people with reduced mobility in an urban environment. However, the design of such a system is constrained by several obstacles. First, we noticed that the vocabulary of the terms coming from these domains was inconstant, and that the whole scientific community is not unanimous on the definition of certain concepts. Our first contribution consists in an ontology, described in OWL, which covers all the elements and concepts necessary for the design of this type of system. Secondly, several current scientific studies underline the lack of a generic architecture model for such systems. As a result, the myriad of solutions proposed in this field suffer from strong limitations in terms of interoperability and reusability of the works. Therefore, our second contribution translates into a generic model facilitating the design of any type of ambient system dedicated to assistance. This model is based on the proposed ontology, while keeping the users at the center of the model. Finally, it has been noticed that ambient systems dedicated to assistance are deployed in an ad-hoc way, either in simulation or in real environment, with an evaluation limited to a statistical analysis of the outputs, or by an evaluation questionnaire submitted to users. We propose an original approach to the implementation of such systems. On the one hand, this approach is based on a simulation-based implementation of the system, on which real elements are progressively added, until a complete deployment in real environment. On the other hand, to evaluate the direct impact of the system, we introduce the analysis of its impact by tools coming from the field of user social network analysis., La transformation des villes en « villes intelligentes » est un des objectifs que les gouvernements visent depuis une dizaine d’années. Cette transformation touche à plusieurs aspects de la ville, dont l’éco-citoyenneté, la gouvernance durable ou encore la mobilité intelligente. C’est sur cet aspect que nos travaux se focalisent. En effet, malgré une certaine progression depuis quelques années, les personnes à mobilité réduite sont souvent oubliées dans la question de l’amélioration des déplacements urbains grâce aux technologies de l’information et de la communication.Parmi ces technologies, l’intelligence artificielle englobe plusieurs domaines et est en plein essor. Un de ces derniers, l’intelligence ambiante vise à la conception de services interconnectés interagissant avec l’environnement physique réel pour offrir des services visant à supporter les activités humaines. C’est dans ce contexte que se placent nos travaux : l’étude et la construction d’un système ambiant dédié à l’assistance des personnes à mobilité réduite en milieu urbain. Or, plusieurs verrous contraignent la conception d’un tel système. Tout d’abord, dans ces domaines, nous avons remarqué que le vocabulaire des termes issus de ces domaines était inconstant, et que toute la communauté scientifique n’est pas unanime sur la définition de certains concepts. Notre première contribution consiste en une ontologie, décrite en OWL, qui recouvre l’ensemble des éléments et concepts nécessaires à la conception de ce type de systèmes. Ensuite, plusieurs études scientifiques actuelles soulignent le manque de modèle d’architecture générique pour de tels systèmes. Ce faisant, la myriade de solutions proposées dans ce domaine souffre de limites fortes sur le plan de l’interopérabilité et de la réutilisabilité des travaux. C’est pourquoi notre deuxième contribution se traduit en un modèle générique facilitant la conception de tout type de système ambiant dédié à l’assistance. Ce modèle s’appuie sur l’ontologie proposée, tout en gardant les utilisateurs au centre du modèle. Enfin, il a été remarqué que les systèmes ambiants dédiés à l'assistance sont déployés de manière ad-hoc. Soit en simulation, en évaluant statistiquement les sorties du système. Soit en milieu réel, où l'évaluation du système se fait par la même observation statistique des sorties, ou par un questionnaire d'évaluation soumis aux utilisateurs. Nous proposons une approche originale de mise en œuvre de tels systèmes. D'une part, cette dernière se base sur une mise en opération du système par simulation hybride, sur laquelle les éléments réels sont ajoutés progressivement, jusqu'à un déploiement en milieu réel complet. D'autre part, dans le but d'évaluer l'impact direct du système, nous proposons l'analyse de son impact par les outils issus du domaine de l'analyse du réseau social des utilisateurs.

Lifetime improvement is a major concern for energy constrained wireless networks. Clustering the network topology is widely utilized for managing and enhancing the system duration. With conventional clustering mechanism Cluster Heads (CHs) close to the Base Station (BS) utilize higher power resource for relaying data packets of the other network CHs. This scenario obstruct the network performance as nodes close to the BS attend an earlier death than their desired durability due to the overloaded routing task. This scenario unbalanced energy consumption and is designated as the hot spot problem. The interest in this work is to carry the intra clustering topology in a vast scale contexts to support the network rising and fairly power balance the energy consuming. In this context, we present an Improved Lifetime Optimization Clustering (ILCK) approach that uses the Kruskal minimal spanning tree heuristic (MST) and consider the state of health (SOH) of devices batteries for the network life maximization. ILCK appeal the Kruskal algorithm in a distributed trend to achieve a minimal MST tree inside wide cluster to consolidate the intra cluster routing topology and mitigate the energy allocated to wireless communications. To the best of our awareness, this is a primary solution that merge the Kruskal approach within an uneven clustering to prolong the objects battery endurance and ease the energy hot spot routing issues. The complexity proof of the proposed approach is provided and simulation results denote that ILCK can adequately scale down the power consumption and lengthen the execution time of the deployed network.

Modular Robots Self Reconfiguration (MRSR) is one of the most challenging problems in nowadays robotics field. This problem consists in the determination of how a set of identical modular robots, with local knowledge of the system and limited energy and computational capacities, can reorganize themselves into a target topology or shape. MRSR has received great attention from the research community. Therefore, a lot of centralized and decentralized algorithms were designed to answer this problem. Unfortunately, the analysis of why and when an algorithm is better than another is less studied. In this paper, we proposed a hybrid centralized/distributed modular robots reconfiguration approach. In this approach, a convolution neural network system is used to estimate the most adapted distributed reconfiguration algorithm according to the initial shape formed by the modular robots and the target shape. Two distributed algorithms are studied: C2SR and TBSR. The designed CNN model allows determining which option is the best for a given reconfiguration problem: use of the C2SR algorithm, use of the TBSR algorithm, or both algorithms are equivalent. The obtained results show that the ML tool succeeds 97.25% of the time to determine the suitable algorithm based on the initial and the final shapes. In addition, the system can be extended to any number of algorithms. Our contribution is the production of a neural network built for the selection of the best modular robots self-reconfiguration algorithm.

How can a set of identical mobile agents coordinate their motions to transform their arrangement from a given starting to a desired goal configuration? We consider this question in the context of actual physical devices called Catoms, which can perform reconfiguration, but need to maintain connectivity at all times to ensure communication and energy supply. We demonstrate and animate algorithmic results, in particular a proof of hardness, as well as an algorithm that guarantees constant stretch for certain classes of arrangements: If mapping the start configuration to the target configuration requires a maximum Manhattan distance of d, then the total duration of our overall schedule is in O(d), which is optimal up to constant factors.

We present a high-voltage-generation-and-multiplexing (HVGM) chip, specifically designed for electrostatic actuation of micro-robots. It can individually control 12 pairs of +/- electrodes using a positive and negative charge pump and mux-structure, consumes 286nW in steady state and 533nW when transitioning a 10pF electrode at 155V/s, and produces a differential voltage of 103V (29× voltage gain from 3.6V) in measurement. We also show a complete microsystem of stacked die, measuring 3×1.4×1.1mm, including HVGM, processor, radio, and harvester for energy autonomous operation.

Energy optimization is an imminent worldwide issues for green computing, it constitutes a major concern and a critical aspect especially for energy constrained wireless networks. To overcome this issue, clustering techniques were introduced as a prominent method that arranges the system operation in correlated manner to attend the energy preservation and prolong the network lifespan. However, existing clustering works only focus on preserving the battery charge to operate until it drains out. This approach is most appropriate for non-rechargeable batteries. However, rechargeable batteries become commonly used and need to be considered. The full discharge of echargeable battery does not mean the device obsolescence. Therefore, the system lifetime optimization should take into consideration the degradation of the rechargeable batteries performances. In this context, we proposed an improved long-term energy efficient unequal clustering approach based on the battery state of health for IoT networks (ILEC_SOH). This work represents an initial step in the integration of the battery health degradation into the unequal network clustering. The obtained results show that the consideration of battery state of health (SOH) significantly improve the network lifespan in the long term compared to the conventional energy efficient approaches.

Based on two types of priority queues, this paper proposes a polling scheduling strategy with reserved channel (Ps-rc strategy) for predefined priority services in cognitive radio networks (CRNs). Channel assembling (CA) technology and spectrum adaptation (SA) technology are adopted to dynamically adjust the assembled channels of secondary users (SUs) to improve the performance of the secondary network. Specifically, the SUs in CRNs are divided into two queues with different priorities; based on polling scheduling, a part of the idle channel is reserved for high-priority queue during the polling stage of the low-priority queue. The purpose is to increase the service quality (QoS) of high priority on the basis of providing fair scheduling. Furthermore, the CA-based channel access process of the proposed strategy is presented and modeled by continuous time Markov chain (CTMC). Then, the process of resource flow between users is mapped on CTMC, and the transition conditions and parameter sets of channel assembling covering all user activities of the system are derived. Finally, the system performance of the proposed CA-based Ps-rc scheduling strategy is simulated and evaluated, including network capacity, spectrum utilization, blocking probability, and forced termination probability. Numerical results show that the proposed strategy can improve the QoS of the predefined high-priority service without causing excessive starvation problem of low-priority service.

As technology improves, new kind of devices are given the ability to interconnect to others, thereby shaping today’s networks. Sensors, smartphones, tablets, laptops, desktops, servers, be they organized or not, altogether form the Internet of Things (IoT), the cloud, and the glue between them. All these devices are heterogeneous in terms of hardware (CPUs, memory type and size, network interfaces, etc), software (Unix, Windows, Android, etc) and behavior (mobility, reliability). However they all have in common the ability to perform computations. But in this context of extreme heterogeneity, distributed computing has become even harder, and the tools we use for clusters, grids and clouds cannot be used. A few middleware solutions to distributed computing this is difficult environment have been proposed recently, but these tools address only a limited number of issues, and they can hardly be coupled together to a single usable solution. As a consequence, Researchers most often opt for developing their own solution, or resorting to simulation. At I3S we also did that. Through several Research projects which implied many aspects of cluster computing and IoT, we built a codebase of successful solutions to many problems we faced along the years. Revamped and generalized into the most consistent, efficient and cleanest possible middleware, we now expose it to the Research communities as Idawi: an Open Source Java message/queue/service/object-oriented decentralized overlay network of components designed to meet the conceptual and practical needs of Researchers working on distributed algorithms.

Tracking objects in indoor is always a challenge for a variety of Internet of Things (IoT) applications. Nowadays, many indoor tracking applications use Low-Power Wide-Area Networks (LPWAN) for Machine-to-Machine (M2M) communication. LoRa is a promising LPWAN radio communication technology designed for wide-area coverage and low-power embedded IoT devices. The objective of this paper is to evaluate the performance of the Angle of Arrival (AoA) direction finding approach in an indoor environment. The AoA performance of the LoRa signal is evaluated using different modulation settings, including Channel Bandwidth (BW) and spreading factor (SF). We measure the AoA accuracy of the LoRa signal using an individual Universal Software Radio Peripheral (USRP) B210, Software Defined Radio (SDR) receiver with the help of the GNU radio software development toolkit. In addition, a new approach to measure the AoA of LoRa signal under interference is investigated. We show that by using the Autocorrelation function combined with our direction finding algorithm, the detection and measurement of two simultaneous receptions is possible. The entire experimental setup was implemented in our indoor office environment. Index Terms-Long-Range (LoRa), Low-Power Wide-Area Networks (LPWANs), Software Defined Radio (SDR), Angle of Arrival (AoA).

Opportunistic software composition of services is a novel interactive approach for the construction of software in open and dynamic ambient environments. The goal is to dynamically provide relevant applications to a user without predefined assembly plan or functional requirements. For that, an intelligent composition system builds, through distributed and interactive reinforcement learning, assemblies of software components present in the user's environment. A current limit of this approach is that in some situations, for example at startup, the composition engine lacks information and as a result proposes random assemblies to the user. The contribution discussed in this paper assists the engine in such situations by adding a feature model generated from the ambient environment. Thus, the engine gathers additional knowledge comparing its proposition to this feature model, providing more pertinent assemblies to the user.

For N ≥ 2, an N-qubit doily is a doily living in the N-qubit symplectic polar space. These doilies are related to operator-based proofs of quantum contextuality. Following and extending the strategy of [SdBHG21] that focused exclusively on three-qubit doilies, we first bring forth several formulas giving the number of both linear and quadratic doilies for any N > 2. Then we present an effective algorithm for the generation of all N-qubit doilies. Using this algorithm for N = 4 and N = 5, we provide a classification of N-qubit doilies in terms of types of observables they feature and number of negative lines they are endowed with.&nbsp

International audience

No study has focused on the economic burden in non-Hodgkin lymphoma (NHL) survivors, even though this knowledge is essential. This study reports on health care resource use and associated health care costs as well as related factors in a series of 1671 French long-term NHL survivors. METHODS: Health care costs were measured from the payer perspective. Only direct medical costs (medical consultations, outpatient treatments, hospitalizations, and medical transport) in the past 12 months were included (reference year 2015). Multiple linear regression was used to search for explanatory factors of health care costs. RESULTS: In total, 1100 survivors (66%) reported having used at least 1 health care resource, and 867 (52%) reported having used at least 1 outpatient treatment. After the authors accounted for missing data, the mean health care cost was estimated at €702 ± €2221. Hospitalizations and outpatient treatments were the main cost drivers. Sensitivity analyses confirmed the robustness of the results. For the 1100 survivors who reported using at least 1 health care resource, the mean health care cost was €1067 ± €2268. Several factors demonstrated statistically significant relationships with health care costs. For instance, cardiovascular disorders increased costs by 66% ± 16%. In contrast, rituximab or autologous stem cell transplantation as initial therapy had no effect on health care costs. CONCLUSIONS: The consideration of economic constraints in health care is now a reality. This retrospective study reports on a better understanding of health care resource use and associated health care costs as well as related factors. It may help health care professionals in their ongoing efforts to design personcentered health care pathways.

International audience; Weighted voting is a conventional approach to improving the performance of replicated systems based on commonly-used majority quorum systems in heterogeneous environments. In long-lived systems, a weight reassignment protocol is required to reassign weights over time in order to accommodate performance variations accordingly. The weight reassignment protocol should be consensus-free in asynchronous failure-prone systems because of the impossibility of solving consensus in such systems. This paper presents an efficient consensus-free weight reassignment protocol for atomic storage systems in heterogeneous, dynamic, and asynchronous messagepassing systems. An experimental evaluation shows that the proposed protocol improves the performance of atomic read/write storage implemented by majority quorum systems compared with previous solutions.

International audience; Development of small hydropower plants depends on its economic and financial feasibility, which must be evaluated by cost estimates before construction. Cost of electromechanical (EM) equipment represents the major portion of the total plant budget. The paper presents new cost estimation approach for EM equipment of hydropower plants. Compared to the literature approaches, the proposed cost correlation is elaborated basing on dataset of hydropower plants located in different continents. Furthermore, the transport cost has been considered of continental factors. Mono-and multi-objective genetic algorithm (GA) optimization have been both applied and compared to determine the best cost correlation. The accuracy of this best cost correlation is validated by using statistical analysis tools to compare it with the equation deduced from the best model reviewed in the literature. The results have shown that the bivariate correlation, viz., the Pearson Product Moment Correlation Coefficient (PPMCC), is slightly improved from 98 % to 99 %. The Mean Square Relative Error (MSRE) is substantially improved from 4.47 % to 1.82 %, and the Utmost Square Relative Error (USRE) from 45.2 % to 11.4 %. Indeed, the proposed cost correlation gives a more accurate estimate of EM equipment cost of hydropower plants in different continents.

Most often, image segmentation is not fully automated and a user is required to lead the process in order to obtain correct results. In a medical context, segmentation can furnish much information to surgeons, but this task is rarely executed. Artificial Intelligence (AI) is a powerful approach for devising a viable solution to fully automated treatment. In this paper, we have focused on kidneys deformed by nephroblastoma. However, a frequent medical constraint is encountered which is a lack of sufficient data with which to train our system. In function of this constraint, two AI approaches were used to segment these structures. First, a Case Based Reasoning (CBR) approach was defined which can enhance the growth of regions for segmentation of deformed kidneys using an adaptation phase to modify coordinates of recovered seeds. This CBR approach was confronted with manual region growing and a Convolutional Neural Network (CNN). The CBR system succeeded in performing the best segmentation for the kidney with a mean Dice of 0.83. Deep Learning was then examined as a possible solution, using the latest performing networks for image segmentation. However, for relevant efficiency, this method requires a large data set. An option would be to manually segment only certain representative slices from a patient and then use them to train a Convolutional Neural Network (CNN) how to segment. In this article the authors propose an evaluation of a CNN for medical image segmentation following different training sets with a variable number of manual segmentations. To choose slices to train the CNN, an Overlearning Vector for Valid Sparse SegmentatIONs (OV 2 ASSION) was used, with the notion of gap between two slices from the training set. This protocol made it possible to obtain reliable segmentations of per patient with a small data set and to determine that only 26% of initial segmented slices are required to obtain a complete segmentation of a patient with a mean Dice of 0.897.

To support users with disabilities, Android provides the accessibility services, which implement means of navigating through an app. According to the Android developer’s guide: “Accessibility services should only be used to assist users with disabilities in using Android devices and apps”. However, developers are free to use this service without any restrictions, giving them critical privileges such as monitoring user input or screen content to capture sensitive information. In this paper, we show that simply enabling the accessibility service leaves 72 % of the top finance a nd 80 % of the top social media apps vulnerable to eavesdropping attacks, leaking sensitive information such as logins and passwords. A combination of several tools and recommendations could mitigate the privacy risks: We introduce an analysis technique that detects most of these issues automatically, e.g. in an app store. We also found that these issues can be automatically fixed in almost all cases; our fixes have b een accepted by 70 % of the surveyed developers. Finally, we designed a notification mechanism which would warn users against possible misuses of the accessibility services; 50 % of users would follow these notifications.

International audience; This work focused on evaluating some machine learning models for their application in e-Health IoT platforms used to detect epileptic seizures. The evaluation is based on two groups of metrics: statistical validation and computational complexity. These metrics determine relevant factors for the models' selection based on the intrinsic efficiency of ML models to detect seizures and their IoT appropriateness to reduce computation and, therefore, energy use. The evaluation scenario defines an EFC architecture with Edge, Fog, and Cloud layers, where the models are deployed initially in the cloud layer for training and validation, to later be deployed in the fog and edge layers for use. Results highlight that GBC and XGBC models present better performance when executed from the cloud; LR, NB and SNN models can be trained from fog nodes, and finally, SLR and MLP can be deployed and used from edge nodes. MLP especially presents a good balance between a low computational cost and a high accuracy in seizure detection.

Modern cities are becoming “green and sustainable” once effective and optimized energy management of resources, gathering, and integration of data from smart energy and building are executed. The integration of heterogeneous technologies, and devices, require an interoperable solution to describe devices and data exchanged.SAREF is an ontology supported by the ETSI SmartM2M standard to achieve interoperability among IoT projects, architectures, etc. that can be extended to any IoT vertical domains such as smart buildings or energy. SAREF can be used to describe data sent through communication protocols or once that data must be processed (e.g., on the cloud, gateways, devices). ETSI does not provide tools yet that support the SAREF ontology to avoid engineers to develop from scratch. We design a SAREF-compliant sensor dictionary which also overcomes SAREF limitations. The sensor dictionary, applied to energy scenarios, is employed by a reasoner to infer meaningful knowledge from sensor data. Online demonstrators are available. The scenarios are also relevant for the Interconnect European-funded project that comprises 50 partners to design interoperable smart building and grids.

International audience; In recent years, there has been an increasing interest in intelligent product especially for industrial applications. A special intelligent product was proposed ten years ago, and this paradigm was named "communicating material", i.e. materials equipped with micro components able to sense, store and communicate data. Communicating material needs a recursive architecture to be managed with a digital twin. A literature analysis is proposed concerning recursive multi-agent systems and architectures regarding communicating material applicability. The most promising reference architecture was adapted for modelling digital twin of a communicating material under energy constraint. The composition and decomposition mechanisms are detailed to improve energy performance by generating an aggregated product with a global view. An applicative scenario relative to the different aggregation modes between two communicating materials, illustrates the proposed architecture ability.

Cyber-physical and ambient systems consist of fixed or mobile devices connected through communication networks. These devices host software components that provide services and may require other services to operate. These software components are usually developed, installed, and activated independently of each other and, with the mobility of users and devices, they may appear or disappear unpredictably. This gives cyber-physical and ambient systems an open and changing character. Software components are bricks that can be assembled to form applications. But, in such a dynamic and open context, component assemblies are difficult to design, maintain and adapt. Applications are used by humans who are at the heart of these systems. Ambient intelligence aims to offer them a personalized environment adapted to the situation, i.e. to provide the right application at the right time, anticipating their needs, which may also vary and evolve over time. To answer these problems, our team is exploring an original approach called "opportunistic software composition", which consists in automatically building applications on the fly from components currently available in the environment, without relying on explicit user needs or predefined assembly plans. In this way, applications emerge from the environment, taking advantage of opportunities as they arise. This thesis defines a software architecture for opportunistic software composition and proposes an intelligent system, called "opportunistic composition engine", in order to automatically build relevant applications, both adapted to the user and to the surrounding environment. The opportunistic composition engine periodically detects the components and their services that are present in the ambient environment, builds assemblies of components, and proposes them to the user. It automatically learns the user's preferences according to the situation in order to maximize user satisfaction over time. Learning is done online by reinforcement. It is decentralized within a multi-agent system in which agents interact via a protocol that supports dynamic service discovery and selection. To learn from and for the user, the latter is put in the loop. In this way, he keeps control over his ambient environment, and decides on the relevance of the emerging application before it is deployed. The solution has been implemented and tested. It works in conjunction with an interface that describes the emerging applications to the user and allows him to edit them. The user's actions on this interface are sources of feedback for the engine and serve as an input to the reinforcement learning mechanism., Les systèmes cyber-physiques et ambiants sont constitués d'appareils fixes ou mobiles reliés par des réseaux de communication. Ces appareils hébergent des composants logiciels qui fournissent des services et peuvent nécessiter d'autres services pour fonctionner. Ces composants logiciels sont généralement développés, installés et activés indépendamment les uns des autres et, avec la mobilité des utilisateurs et des appareils, ils peuvent apparaître ou disparaître avec une dynamique imprévisible. Ceci donne aux systèmes cyber-physiques et ambiants une nature ouverte et changeante. Les composants logiciels sont des briques que l'on peut assembler pour former des applications mais, dans un tel contexte de dynamique et d'ouverture, les assemblages de composants sont difficiles à concevoir, à maintenir et à adapter. Les applications sont utilisées par des humains qui sont donc au cœur de ces systèmes. L'intelligence ambiante vise à leur offrir un environnement personnalisé adapté à la situation, c'est-à-dire à fournir la bonne application au bon moment, en anticipant leurs besoins, qui peuvent aussi changer dans le temps. Pour répondre à ces problèmes, notre équipe explore une approche originale appelée "composition logicielle opportuniste" qui consiste à construire automatiquement des applications à la volée à partir des composants disponibles sur le moment dans l'environnement, sans s'appuyer sur des besoins explicites de l'utilisateur ni sur des plans d'assemblage prédéfinis. Ainsi, les applications émergent de l'environnement, en tirant parti des opportunités au fur et à mesure qu'elles se présentent. Cette thèse définit une architecture logicielle pour la composition logicielle opportuniste et propose un système intelligent, appelé "moteur" de composition opportuniste, afin de construire automatiquement des applications pertinentes, à la fois adaptées à l'utilisateur et à l'environnement ambiant. Le moteur de composition opportuniste détecte périodiquement les composants et leurs services présents dans l'environnement ambiant, construit des assemblages de composants et les propose à l'utilisateur. Il apprend automatiquement les préférences de l'utilisateur en fonction de la situation afin de maximiser ultérieurement sa satisfaction. L'apprentissage se fait en ligne par renforcement. Il est décentralisé au sein d'un système multi-agent dans lequel les agents interagissent via un protocole qui prend en charge la découverte et la sélection dynamique de services. Pour apprendre de l'utilisateur et pour l'utilisateur, ce dernier est mis dans la boucle. Ainsi, il garde le contrôle sur son environnement ambiant, et décide de la pertinence de l'application émergente avant qu'elle ne soit déployée. La solution a été implémentée et expérimentée. Elle fonctionne de manière couplée avec une interface qui décrit les applications émergentes à l'utilisateur et lui permet de les modifier. Les actions de l'utilisateur sur cette interface sont sources de feedback pour le moteur et servent à alimenter le mécanisme d'apprentissage par renforcement.

International audience; One of the major failures attributed to pipeline transportation of crude oil is oil leakages and spills. Hence, it is monitored via several classical leak detection techniques (LDTs), which are more recently implemented on centralised wireless sensor networks (WSN)-based leak detection and monitoring systems (LDMS). However, the LDTs are sometimes prone to high false alarms, and the LDMS are sensitive to single points of failure. Thus, we propose HyDiLLEch, a distributed leakage detection and localisation technique based on a fusion of several LDTs. In this work, we implemented HyDiLLEch and compared it to the individual LDTs in terms of communication efficiency and leakage detection and localisation accuracy. With HyDiLLEch, the number of nodes detecting and localising leakages increases by a maximum of four to six times, thereby eliminating single points of failures. In addition, we improve the accuracy of localisation in nodes physically-close to the leak and maintain an average of 96% accuracy with little to no communication overhead.

The concept of cashier-free shopping started to become popular since 2016. Its goal is to enable customers to avoid checking out at cashiers to save both the consumer's time and the shop owner's cost on staff. Recently there are many commercialized samples deployed worldwide. However, their technology is patented and the key performance is not yet public. Therefore, it is important to analyze their underlying technology from a research perspective to help people understand this area and related work. This survey summarizes existing research work on cashier-free shopping systems and other papers on supporting techniques.

Information Systems are evolving, integrating more and more mobile & heterogenous resources, in addition to those traditionally available on data centers and cloud platforms. Among the current trends leading to this new generation, called Pervasive Information Systems, we may cite the fog computing. This paradigm promotes an opportunistic use of devices on the edge of the network, by bringing computation nearest the user and the data production. Nevertheless, heterogeneity of fog environments complexify the resource management, and notably the task scheduling. In this paper, we tackle this issue by proposing a conceptual architecture supporting an opportunistic resource management of fog resources on PIS. The cornerstone of the solution is the use of context information for (i) guiding the tasks scheduling on heterogeneous & dynamic environment; and (ii) expressing the resources & services policies conditioning their use.

Le présent document synthétise plusieurs travaux de recherche réalisés entre 2002 et 2020, traitant de la notion de contexte sous différents angles. Ces différentes contributions, croisant plusieurs communautés de l’Informatique (CSCW, Ubiquitous Computing, Systèmes d’Information), sont ici discutées, à travers les problématiques abordées et les propositions qui ont été faites au cours des années. Toutes ces contributions partagent une ligne directrice commune, la notion de contexte, qui a été appliquée tout au long de ces contributions sur différents types de systèmes, notamment les collecticiels (systèmes de Groupware) et les Systèmes d'Information. Toutes ces contributions conduisent à une perspective commune, sur une nouvelle génération de Systèmes d'Information, appelée ici Systèmes d'Information Pervasifs, dans laquelle la notion de contexte joue un rôle clé à des fins d'adaptation. Chacune des contributions présentées dans ce document converge ainsi vers cette nouvelle génération de SI, et vers une vision d’une « context facility », dans laquelle le support à la notion de contexte pourrait être vu comme un service offert par les SIP à ses différents composants, quel que soit leur niveau.

Understanding and predicting how humans move within space and time is of fundamental importance for many scientific domains, such as epidemic propagation (e.g., the COVID-19 pandemics), mobile networks, urban planning (e.g., anticipatory temperature control of a home for energy consumption management), or ride-sharing. Yet, apprehending human mobility is intrinsically complex. On the one hand, human movements are constrained by physical presence in workplaces, gyms, or universities, in addition to the involvement in routine and social activities. On the other hand, the large variety of leisure places and the availability of modern means of transportation allow people to break their routinary patterns to discover new places. Understanding human mobility is a longstanding challenge that goes back to the 19th century. Up to the 20th century, scholars focused only on group and migration flows in view of the quality of the leveraged data (Census data or surveys). But recently, with the ubiquity of mobile devices, Internet connectivity, and positioning systems (e.g., GPS system), capturing individuals' daily whereabouts at very fine spatial and temporal scales has become possible. This offers the opportunity to observe and study human mobility at the individual level with an unprecedented level of detail. In particular, the increasing availability of such data has led to ongoing development in the field of mobility research. Nevertheless, the scientific literature on human mobility prediction is oblivious to individuals' tendencies for novelty-seeking, i.e., exploring and discovering new places. Conventional predictors relying on personal geographical data perform poorly when it comes to discoveries of new regions. The reason is explained by the prediction relying only on previously visited/seen (or known) locations. As a side effect, places that were never visited before (or explorations) by a user cause disturbance to known location's prediction. Neglecting novelty-seeking activities at first glance appears to be inconsequential on the ability to understand and predict individuals' trajectories. In this manuscript, we claim and show the opposite: exploration-like visits strongly impact mobility understanding and anticipation.This thesis focuses on exploratory visits in human mobility. It first seeks to unveil the exploratory preferences of the population. Afterward, it evaluates the impacts of exploration-like visits on the theoretical and practical predictability extents. Finally, it proposes an exploration-aware mobility prediction framework that integrates the notion of exploration.Throughout this manuscript, we reveal the existence of three distinct mobility profiles with regard to the exploration activity - Scouters (i.e., extreme explorers who are keener to discover new places), Routiners (i.e., extreme returners who limit their mobility to a few locations), and Regulars (i.e., without extreme behavior). Besides, we show that the novelty-seeking factor is an essential element to consider and should not be overlooked when designing predictors, particularly for specific categories exhibiting high exploration activities. Furthermore, by integrating the notion of exploration in prediction, we demonstrate substantial improvements in prediction accuracy by dint of fruitfully forecasting coarse-grained zones used for exploration activities., Comprendre et prédire comment les humains se déplacent dans l'espace et le temps est d'une importance fondamentale pour de nombreux domaines scientifiques tels que la propagation des épidémies (par exemple, la pandémie du COVID-19), les réseaux mobiles, l'urbanisme (par exemple, le contrôle anticipé de la température d'une maison pour la gestion de la consommation énergietique), ou le covoiturage. Or, appréhender la mobilité humaine est intrinsèquement complexe. D'une part, les mouvements humains sont limités par l’obligation de présence physique sur les lieux de travail, les gymnases ou les universités en plus de la participation à des activités routinières et sociales. D'autre part, la grande variété de lieux de loisirs et la disponibilité de moyens de transport modernes permettent aux individus d’interrompre leurs routines pour découvrir de nouveaux lieux. Comprendre la mobilité humaine est un défi de longue date qui remonte au 19 e siècle. Jusqu'au 20 e siècle, les chercheurs se sont concentrés uniquement sur les flux collectifs et migratoires au vu de la qualité des données exploitées (données de recensement ou d'enquêtes). Mais récemment, avec l'omniprésence des appareils mobiles, de la connectivité Internet et des systèmes de positionnement (par exemple, le système GPS), il est devenu possible de capturer les allées et venues quotidiennes des individus à des échelles spatiales et temporelles très fines. Cela offre la possibilité d'observer et d'étudier la mobilité humaine au niveau individuel avec un niveau de détail sans précédent. En particulier, la disponibilité croissante de ces données a conduit à un développement continu dans le domaine de la recherche sur la mobilité. Néanmoins, la littérature scientifique sur la prédiction de la mobilité humaine ignore les tendances des individus à rechercher la nouveauté, c'est-à-dire à explorer et à découvrir de nouveaux endroits. Les prédicteurs conventionnels reposant sur des données géographiques personnelles fonctionnent mal lorsqu'il s'agit de découvrir de nouvelles régions. La raison est expliquée par la prédiction reposant uniquement sur des emplacements précédemment visités/vus (ou connus). Comme effet secondaire, des emplacements qui n'ont jamais été visités auparavant (ou des explorations) par un utilisateur perturbent la prédiction d'un emplacement connu. Négliger à première vue les activités de recherche de nouveauté apparaît sans conséquence sur la capacité à comprendre et à prévoir les trajectoires des individus. Dans ce manuscrit, nous affirmons et montrons le contraire : les visites de type exploration impactent fortement la compréhension et l'anticipation de la mobilité.Cette thèse porte sur les visites exploratoires en mobilité humaine. Elle cherche d'abord à dévoiler les préférences exploratoires de la population. Ensuite, elle évalue les impacts des visites de type exploration sur les étendues de prévisibilité théorique et pratique. Enfin, elle propose un prédicteur de mobilité qui intègre la notion d'exploration.À travers le manuscrit, nous révélons l'existence de trois profils de mobilité distincts en ce qui concerne l'activité d'exploration - Scouters (c'est-à-dire les explorateurs extrêmes qui sont plus enclins à découvrir de nouveaux endroits), Routiners (c'est-à-dire les routiniers extrêmes qui limitent leur mobilité à quelques endroits), et Regulars (c'est-à-dire sans comportement extrême). En outre, nous montrons que le facteur de recherche de nouveauté est un élément essentiel à considérer et ne doit pas être négligé lors de la conception de prédicteurs, en particulier pour des catégories spécifiques d’individus présentant des activités d'exploration élevées. De plus, en intégrant la notion d'exploration dans la prédiction, nous démontrons des améliorations considérables de la précision de la prédiction grâce à la prévision fructueuse des zones à gros grains utilisées pour les activités d'exploration.

International audience

International audience

With the popularization of the Internet of Things, more and more intelligent products appear in our daily life. The McBIM project (Materials communicating with the Building Information Modelling) aims to design a new type of “communicating concrete”: concrete elements equipped with embedded wireless sensor network (WSN). This communicating concrete can measure the physical environment, store the information and exchange data with BIM platforms. One objective is to demonstrate the usefulness of the concept over the whole building lifecycle and the energy management is one of the facing challenges. Meanwhile, the product, user information as well as the large amount of monitoring data bring challenges regarding data management. To solve both issues, this work proposes to couple the physical communicating concrete to an external, digital and intelligent system in order to extend network lifetime. In this thesis, these physical and digital parts are both studied. In the physical part, energy saving methods are reviewed, to understand how to extend the network lifetime. Network routing, in-network data processing and time synchronization are considered to build an energy model for periodic data collection. This energy model is validated with an Arduino-XBee based platform. The experiment shows that the analytic energy estimation model provides an approximate but accurate remaining energy level. It can be used online by nodes or remotely by user to the remaining energy. Energy-related data and collection modes can be managed by the digital part, based on a multi-agent system (MAS). The MAS can not only store monitoring data of communicating elements but also provide energy efficient data collection or organization strategies. Composition of communicating elements may happen in construction or exploitation lifecycle’s phase to reduce energy consumption. A recursive multi-agent architecture is proposed to meet high flexibility requirements. This multi-agent system is implemented on a 3D simulation platform to visualize the communicating concrete and verify the usefulness of our analytic energy model. The recursive multi-agent architecture provides energy efficiency solutions to extend network lifetime and respond to environmental changes., Avec la popularisation de l’Internet des Objets, de plus en plus de produits intelligents apparaissent dans notre vie. Le projet McBIM (Matière communicante au service du BIM-Building Information Modelling) vise à concevoir un nouveau type de « béton communicant » qui est le béton équipé d'un réseau de capteurs sans fil embarqué (RSCF). Ce béton communicant est capable de mesurer des phénomènes physiques, de stocker des informations et d'échanger des données avec des plateformes BIM. Pour démontrer l'utilité du concept sur l'ensemble du cycle de vie, l'exigence de service à long terme pose des défis relatifs à la durée de vie du réseau intégré dans la matière. Dans le même temps, la grande quantité de données de surveillance posent des défis relatifs à la gestion d'informations. Dans ces travaux, nous proposons de coupler le béton communicant physique à sa représentation digitale. Ce système externe et intelligent doit pouvoir permettre à terme de prolonger la durée de vie du réseau. La partie physique étudie des méthodes d'économie d'énergie applicables à la collecte de données dans les réseaux de capteurs, afin de comprendre comment prolonger la durée de vie de réseau et construire un modèle de consommation d'énergie du réseau. Le routage du réseau, le traitement des données dans le réseau (In-Network Processing) et la synchronisation des nœuds sont pris en compte pour construire un modèle de consommation énergétique du réseau embarqué, modèle validé avec une plate-forme Arduino-XBee. L'expérience montre que le modèle d'estimation de l'énergie fournit une approximation précise de l'énergie restante, pouvant être utilisée en ligne par les nœuds ou à distance par un utilisateur. La gestion de l'énergie du réseau et des modes de collecte peut-être réalisée par un système multi-agents dédié. La partie digitale est construite avec un système multi-agents, pouvant non seulement stocker les données de surveillance, mais également fournir des stratégies de collecte de données. La composition des éléments communicants peut intervenir en phase de construction ou d'exploitation. Une architecture multi-agents récursive est proposée pour répondre à cette exigence de flexibilité. Ce système multi-agents est implémenté sur une plateforme de simulation 3D pour visualiser le béton communicant et vérifier la validité de notre modèle énergétique analytique. L'architecture multi-agents récursive fournit des solutions efficaces pour estimer l'énergie résiduelle dans la matière, afin de prolonger la durée de vie du réseau et répondre aux changements environnementaux.

International audience

International audience; This paper studies both the design and the cyclic scheduling of multi-hoist treatment surface facilities. Former research have always assumed that the design of the production line is already available with a fixed material handling fleet size. However, in this study, the number of hoists is considered as a decision variable of the problem. The latter is then bi-objective and called the Cyclic Hoist Design and Scheduling Problem (CHDSP). The optimization objective is to determine an optimal cyclic schedule for each possible number of hoists that minimizes the cycle time and thus maximizes the line throughput rate. The achieved results will contribute to building a decision support system that will enable to choose the required number of transportation resources for the production line regarding both aims of productivity maximization and cost savings. An original encoding approach is proposed that both assigns a number of hoists to the line and generates their move sequences, which are evaluated thanks to a mixed integer linear programming model. A rich and well-structured solving algorithm based on variable neighborhood search is developed and adapted to solve efficiently the whole problem. It is also improved by a backtrack procedure that further enhances the findings. Computational experiments are conducted on benchmark problems and demonstrate the high performance and effectiveness of the proposed algorithm that was, in most cases, even able to reach the optimal solutions.

For several years, we have been witnessing a convergence between cyber-physical systems (CPS) and the Internet of Things (IoT). CPS integrate embedded systems with their physical and human environment by ensuring communication between different sensors and actuators. The IoT targets the network and communication protocols between connected objects. This convergence offers prospects for various applications ranging from connected vehicles to smart grids and the factories of the future. The aim of this thesis is to ensure and guarantee the operational safety of CPS-IoT systems. For this, we have considered a specific case study throughout the thesis which is UAVs. Initially, we focused on the different methods of analysis of operational safety that already exist. These methods have proved their worth for the design and implementation of on-board systems. Throughout this process, we tried to answer the following question: are these existing methods adequate to perform the necessary safety analyses for CPS-IoT? It was concluded that new approaches to analyse the safety of operation of CPS-IoT systems are needed due to the significant complexity of these systems. As a second step, a methodology for predictive analysis of the resilience of CPS-IoTs was proposed. Resilience is defined as being the ability of a system to tolerate failures, to continue to provide the requested service while considering the various internal and external constraints of the system. Two different types of resilience have been differentiated: endogenous and exogenous resilience. Endogenous resilience is the inherent ability of the system to detect and deal with internal faults and malicious attacks. Exogenous resilience is the ongoing ability of the system to maintain safe operation in its surrounding environment. The last part of our work was to investigate the impact of artificial intelligence on the safe operation of CPS-IoTs. More specifically, we looked at how artificial intelligence could be used to enhance UAV safety in the path planning phase. The results obtained were compared with existing planning algorithms., Depuis plusieurs années, nous assistons à une convergence entre les systèmes cyber-physiques (CPS) et l’Internet des Objets (IoT). Les CPS intègrent les systèmes embarqués avec leur environnement physique et humain en assurant une communication entre différents capteurs et actionneurs. L’IoT vise le réseau et les protocoles de communication entre les objets connectés. Cette convergence offre des perspectives d’applications diverses allant des véhicules connectés aux réseaux électriques intelligents ainsi qu’aux usines du futur. Le but de cette thèse est d’assurer et garantir la sûreté de fonctionnement des systèmes CPS-IoT. Pour ceci, nous avons considéré un cas d’étude spécifique tout au long de la thèse qui est les drones. Dans un premier temps, on s’est focalisé sur les différentes méthodes d’analyse de sûreté de fonctionnement qui sont déjà existantes. Ces méthodes ont fait leurs preuves pour la conception et la réalisation des systèmes embarqués. Tout au long de ce process, on a essayé de répondre à la question suivante: est-ce que ces méthodes existantes sont adéquates pour réaliser les analyses de sûreté de fonctionnement nécessaires pour les CPS-IoT? On a conclu la nécessité de nouvelles approches pour analyser la sûreté de fonctionnement des systèmes CPS-IoT du fait de la complexité significative de ces systèmes. Dans un second temps, on a proposé une méthodologie pour l’analyse prédictive de la résilience des CPS-IoT. La résilience est définie comme étant la capacité d’un système à tolérer les pannes, à continuer à fournir le service demandé tout en considérant les différentes contraintes internes et externes au système. On a différencié deux types différents de résilience qui sont la résilience endogène et exogène. La résilience endogène est la capacité inhérente du système à détecter et à traiter les défauts internes et les attaques malveillantes. La résilience exogène est la capacité permanente du système à maintenir un fonctionnement sûr dans son environnement ambiant. La dernière partie de notre travail a consisté à investiguer l’impact de l’intelligence artificielle sur la sûreté de fonctionnement des CPS-IoT. Plus spécifiquement, on s’est intéressé à comment serait-il possible d’utiliser l’intelligence artificielle pour accroître la sûreté des drones lors de la phase de planification de chemin. Les résultats obtenus ont été comparés avec les algorithmes de planification existants.

International audience; Information and communication technologies, like Wireless Sensor Network (WSN), allow to imagine new objects. Objects whose material could be composed of micro-elements able of data sensing, storing, processing, and communicating. Communicating Materials (CM) is a new paradigm inspired by Intelligent Product concept. This paper presents a recursive Multi-Agent Framework proposed for handling composition of Communicating Materials, which is an application of the Janus Effect. The proposed framework, still in development, is illustrated on the McBIM project, consisting in designing a communicating concrete and digital services that can be used by Building Information Modelling (BIM) applications throughout its lifecycle. MAS principles are demonstrated on this construction industry project to control WSN energy and to illustrate CM aggregation ability.

Autonomic systems manage resources to achieve high-level goals with minimum human intervention, by continuously adapting based on a model of themselves and of their environment. Adaptation requires managing runtime uncertainties produced by changes within a dynamic environment, by modeling unexpected situations that were not considered at design time. This challenge becomes even more critical in a self-integration context where autonomic systems, traditionally engineered systems, and human organizations, influence each other in a shared environment. In this paper, we propose a cognitive control system that aims to manage runtime uncertainty in self-integrating autonomic systems. The cognitive control system autonomously extends the existing modeling language, by generating knowledge to represent unexpected situations as they are encountered at runtime, and reasons on the resulting model, in order to determine the required adaptation knowledge. As proof-of-concept validation, we present a scenario supported by an implementation of our proposal, illustrating the capabilities of the cognitive control system in the smart home domain. The obtained results set the basis for further studying the modeling language for describing the adaptation knowledge, with the objective of enriching the collaboration capabilities of autonomic systems in a self-integration context as future work.

Mobile edge computing (MEC) allows for the execution of delay-sensitive and context-aware applications close to the end-users while alleviating backhaul utilization and computation at the core network. A recent trend is to extend the capabilities of the MEC using the available resources of end-user devices beyond the edge, such as vehicles. Because of the highly mobile nature of such devices, the beyond-the-edge computing resources available at the time the MEC offloads a task decays over time. In this paper, we propose and investigate the persistence of groups of computational nodes in a vehicular-augmented MEC scenario. The groups are formed according to the time of arrival of vehicles and their co-location within a cell. Through the analyses of two real-world vehicular scenarios in the cities of Rome and Rio de Janeiro. On the one hand, our results show that the persistence is highly heterogeneous, depending on the vehicular traffic density and the time of day. On the other hand, we are able to identify periods during which the persistence is short and periods that show stronger stability – it becomes then possible to schedule tasks to specific time slots depending on the requirements and the expected persistence.

Mobile crowdsensing is a powerful mechanism to contribute to the ubiquitous sensing of data at a relatively low cost. With mobile crowdsensing, people provide valuable observations across time and space using sensors embedded in/connected to their smart devices, e.g., smartphones. Particularly, opportunistic crowdsensing empowers citizens to sense objective phenomena at an urban and fine-grained scale, leveraging an application running in the background. Still, crowdsensing faces challenges: The relevance of the provided measurements depends on the adequacy of the sensing context with respect to the phenomenon that is analyzed; The uncontrolled collection of massive data leads to low sensing quality and high resource consumption on devices; Crowdsensing at scale also involves significant communication, computation, and financial costs due to the dependence on the cloud for the post-processing of raw sensing data. This thesis aims to establish opportunistic crowdsensing as a reliable means of environmental monitoring. We advocate enforcing the cost-effective collection of high-quality data and inference of the physical phenomena at the end device. To this end, our research focuses on defining a set of protocols that together implement collaborative crowdsensing at the edge, combining : Inference of the crowdsensor’s physical context characterizing the gathered data: We assess the context beyond geographical position. We introduce an online learning approach running on the device to overcome the diversity of the classification performance due to the heterogeneity of the crowdsensors. We specifically introduce a hierarchical algorithm for context inference that requires little feedback from users, while increasing the inference accuracy per user. Context-aware grouping of crowdsensors to share the workload and support selective sensing: We introduce an ad hoc collaboration strategy, which groups co-located crowdsensors together, and assigns them various roles according to their respective contexts. Evaluation results show that the overall resource consumption due to crowdsensing is reduced, and the data quality is enhanced, compared to the cloudcentric architecture. Data aggregation on the move to enhance the knowledge transferred to the cloud: We introduce a distributed interpolation-mediated aggregation approach running on the end device. We model interpolation as a tensor completion problem and propose tensor-wise aggregation, which is performed when crowdsensors encounter. Evaluation results show significant savings in terms of cellular communication, cloud computing, and, therefore, financial costs, while the overall data accuracy remains comparable to the cloud-centric approach. In summary, the proposed collaborative crowdsensing approach reduces the costs at both the end device and the cloud, while increasing the overall data quality., Le crowdsensing mobile permet d’obtenir des données sur l’environnement à un coût relativement faible. De fait, les personnes peuvent collecter et partager des observations spatio-temporelles au moyen de capteurs intégrés dans les appareils intelligents comme les smartphones. En particulier, le crowdsensing opportuniste permet aux citoyens de détecter des phénomènes environnementaux à l’échelle urbaine grâce à une application dédiée s’exécutant en arrière plan. Cependant, le crowdsensing est confronté à différents défis : la pertinence des mesures fournies dépend de l’adéquation entre le contexte de détection et le phénomène analysé ; la collecte incontrôlée de données entraîne une faible qualité de détection ainsi qu’une forte consommation des ressources au niveau des appareils ; le crowdsensing à large échelle induit des coûts importants de communication, de calcul et financiers en raison de la dépendance au cloud pour le traitement des données brutes. Notre thèse vise à rendre le crowdsensing opportuniste comme un moyen fiable d’observation de l’environnement urbain. Pour ce faire, nous préconisons de favoriser la collecte et l’inférence du phénomène physique au plus proche de la source. À cet effet, notre recherche se concentre sur la définition d’un ensemble de protocoles complémentaires, qui mettent en œuvre un crowdsensing collaboratif entre les nœuds mobiles en combinant : L’inférence du contexte physique du crowdsensor, qui caractérise les données recueillies. Nous évaluons un contexte qui ne se limite par à une simple position géographique et nous introduisons une technique d’apprentissage du contexte qui s’effectue au niveau de l’appareil afin de pallier l’impact de l’hétérogénéité des crowdsensors sur la classification. Nous introduisons spécifiquement un algorithme hiérarchique pour l’inférence du contexte qui limite les interactions avec l’usager, tout en augmentant la précision de l’inférence. Le groupement contextuel des crowdsensors de manière à partager la charge et effectuer une captation sélective. Nous introduisons une stratégie de collaboration ad hoc qui vise à affecter différents rôles aux crowdsensors, afin de répartir la charge entre les crowdsensors proches en fonction de leur contexte respectif. L’évaluation de notre solution montre une réduction de la consommation globale des ressources et une meilleure qualité des données, comparée à celle basée sur une architecture cloud. L’agrégation nomade de données pour améliorer les connaissances transférées au cloud. Nous introduisons une solution à l’agrégation des données observées qui distribue l’interpolation sur les crowdsensors. Les résultats de son évaluation montrent des gains importants en termes de communication cellulaire, de calcul sur le cloud et donc de coûts financiers, tandis que la précision globale des données reste comparable à une approche centralisée. En résumé, l’approche collaborative proposée pour le crowdsensing réduit les coûts à la fois sur le terminal et sur le cloud, tout en augmentant la qualité globale des données.

International audience

International audience; The growth of the Internet of Things (IoT) results in a proliferation of different protocols (ZigBee, Bluetooth, 6LowPAN, Z-Wave, Wi-Fi, etc.). Organizations tend to quickly deploy several IoT applications over time and thus face heterogeneous IoT systems, combining different IoT protocols in different places of the overall system. This heterogeneity of protocols makes these networks hard to monitor or control, and some misconfigurations or unexpected device behaviours may even expose users to security issues. In this work, we propose IoTMap. IoTMap provides an iterative graph-based modelling of interconnected IoT with a multi-layers view ranging from point-to-point transmissions to application-type analysis. With this modelling, we aim to highlight the network state at a specific time. It can also be used as the intelligence gathering phase during a penetration testing to evaluate the global network security.

International audience; This paper considers generalizing context reasoning capabilities through a context mining facility offered to all Information System applications. This facility requires mining context data at the system scale, which raises several challenges for Machine Learning approaches used for such mining. Through a detailed literature review, we analyze these approaches with regard to the requirements of such a context mining facility at the Information System level, pointing to the potential and to the challenges raised by this perspective.