Publications

2020

• Resilience by design & failures forecasting for a connected autonomous vehicle
  
  • Monteuuis Jean-Philippe
  , 2020. Autonomous vehicles with an automation level 5 will drive autonomously in any road scenarios such as highways, snowy roads, urban areas, or traffic jams. The integration of V2X communication, as a new source of perception for the vehicle could remove the limitations of local perception by communicating with an occluded pedestrian or by detecting in advance the presence of a vehicle under a heavy mist. However, this V2X communication may be a new source of attacks threatening the vehicle perception. Current countermeasures are not designed for all autonomous vehicles because these countermeasures require the driver assistance or work with a specific set of sensors. Therefore, the thesis aims to propose a generic failure resilient perception architecture for all types of connected and autonomous vehicles supporting different kinds of sensors. In this thesis, we propose a generic perception architecture named GPA with its failure resilient perception algorithm (FRPA). We propose a new threat analysis and risk assessment method named SARA that identifies and assess the risk of attacks targeting connected and automated vehicles with an automation level 5. To identify where and how these attacks occur, we propose an attacker and a security goal model for all automotive perception systems. We implemented two modules of our failures resilient perception algorithm (FRPA): a Machine Learning based Failure Classifier and a V2X-Sensor Correlation Module considering three kinds of source: camera, radar, and V2X. We highlighted several new attacks in the perception pipeline and raise the need for new security countermeasures such as the physical integrity of road infrastructures and trustworthy perception algorithms. Besides, our countermeasures based on machine learning and sensor correlation showed very accurate results to detect and classifies perception failures (over 90% accuracy score). Finally, the ideas developed in the thesis resulted in 10 filled patents and several publications.
• Full-duplex for cellular networks : a stochastic geometry approach
  
  • Arrano Scharager Hernan
  , 2020. Full-duplex (FD) is a principle in which a transceiver can receive and transmit on the same time-frequency radio resource. The principle was long held as impractical due to the high self-interference that arises when simultaneously transmitting and receiving in the same resource block. When assuming perfect self-interference cancellation, FD can potentially double the spectral efficiency (SE) of a given point-to-point communication. In practice though, it is not possible to achieve the aforementioned characteristic. Moreover, under a cellular network context, not only the self-interference limits the performance, since additional co-channel interference is created by base stations (BSs) and users equipment (UEs). However, even with the higher interference dowlinks (DLs) still obtain higher SE performances, whereas uplinks (ULs) are generally critically degraded, when compared to half-duplex (HD). We focus our work in the study of alternatives that can help improve the impaired ULs in FD networks, while still trying to profit from the gains experienced by DLs.In this regard, we use stochastic geometry along the thesis as a means to characterize key performance indicators of cellular networks, such as: coverage probability, average SE and data rates. The thesis is divided into three major studies. Firstly, we propose a duplex-switching policy which enables BSs to operate in FD- or HD- depending on the UL and DL conditions. Secondly, we investigate the performance of hybrid HD/FD networks under a millimeter wave context. Finally, we propose a novel algorithm based on nonorthogonal multiple-access (NOMA) and successive interference cancellation (SIC), which allows BSs to coordinate on their respective transmission schemes to reduce the BS-to-BS interference. We demonstrate that the models presented in the thesis allow to balance the gains of one link over the other; reducing the UL degradation, while maintaining DL gains. In addition, we show that scenarios in which equipment is able to perform beamforming are ideal for FD deployments, since they directly reduce the cochannel interference.
• A New Network Configuration Management Architecture for Future Aircraft Systems
  
  • Delmas Thibault
  • Iannone Luigi
  • Garcia Jean-Pierre
  • Monsuez Bruno
  , 2020. Aircraft systems are evolving and being enhanced thanks to new design paradigms leveraging on recent technology advances in embedded systems. However, the Integrated Modular Avionics (IMA) model, used in current avionics, has shown important limitations to accommodate such evolution. These new paradigms demand for much more global system modularity than what IMA is able to offer. Such system evolution has as well an impact on the underlying different networks present on aircrafts. In this context, it is mandatory to investigate the kind and the breadth of adaptation networks need in order to cope with new requirements. To this end, this paper we firstly investigate the current aircrafts network configuration and management procedures. It appears that they lack the features, more specifically, the configuration management features, necessary to support these new use cases. We then look at proposals trying to fulfil this features gap. Each of them, while providing parts of the answers, also come with trade-off or insufficiencies that prevent them from fully answering to the new needs. Then, a new network configuration architecture able to cope with the newly defined configuration management requirements is provided. A comparison to the other approaches is presented, so to highlight how the proposed architecture better fulfils long-term evolution requirements while being less complex and more suitable for current configuration procedures than the other proposal. Finally, a simulation of the configuration architecture is done to provide insights on the new proposed features.
• Hardware / Software / Analog System Partitioning with SysML and SystemC-AMS
  
  • Genius Daniela
  • Apvrille Ludovic
  , 2020. Model-driven approaches for designing software and hardware parts of embedded systems are generally limited to their digital parts. On the other hand, virtual prototyping and co-simulation have emerged as a promising research topic, but target the modeling levels when partitioning has already been performed. This paper presents a model-driven platform for the partitioning of analog/mixed-signal systems. keywords: virtual prototyping, embedded systems , analog/mixed signal, design space exploration
• Towards Formal Verification of Autonomous Driving Supervisor Functions
  
  • Assioua Yasmine
  • Ameur-Boulifa Rabéa
  • Guitton-Ouhamou Patricia
  , 2020. In the software development lifecycle, errors and flaws can be introduced in the different phases and lead to failures. Establishing a set of functional requirements helps producing safe software. However, ensuring that the (being) developed software is compliant with those requirements is a challenging task due to the lack of automatic and formal means to lead this verification. In this paper, we present our approach that aims at analysing a collection of automotive requirements by using formal methods. The proposed approach for formal verification is evaluated by the application to supervisor functions of the autonomous driving (AD) system, the system in charge of self-driving.
• Invited Lecture on Continous-Variable Quantum Key Distribution
  
  • Alleaume Romain
  , 2020.
• Engineering Railway Systems with an Architecture-Centric Process Supported by AADL and ALISA: an Experience Report
  
  • Crisafulli Paolo
  • Blouin Dominique
  • Caron Françoise
  • Maxim Cristian
  , 2020. The increasing automation of transportation systems has contributed to the emergence of the so-called Cyber-Physical Systems (CPS), which are computation-based systems in which computing devices, sensors, actuators and networks collaborate to monitor and control physical entities via feedback loops. To cope with the increasing complexity of such systems, engineering teams require model-based tools, because they can provide early virtual integration and verification, reuse of existing models, requirements traceability and support of an incremental development process. However, these benefits can only be earned if the chosen modelling languages are expressive enough to capture all aspects necessary to perform the virtual verifications with the required confidence degree.
• Prédiction conformelle profonde pour des modèles robustes
  
  • Messoudi Soundouss
  • Rousseau Sylvain
  • Destercke Sébastien
  , 2020, RNTI-E-36, pp.301-308. Les réseaux profonds, comme d'autres modèles, peuvent associer une confiance élevée à des prédictions peu fiables. Rendre ces modèles robustes et fiables est donc essentiel, surtout pour les décisions critiques. Ce papier montre expérimentalement que la prédiction conformelle, et plus particulièrement l'ap-proche de [Hechtlinger et al. (2018)], apporte une solution convaincante à ce défi. La prédiction conformelle fournit un ensemble de classes couvrant la vraie classe avec avec une fréquence choisie au préalable par l'utilisateur. Dans le cas où l'exemple à prédire est atypique, la prédiction conformelle prédira l'en-semble vide. Les expériences menées montrent le bon comportement de l'ap-proche conformelle, en particulier lorsque les données sont bruitées.
• Innovative ATMEGA8 Microcontroler Static Authentication Based on SRAM PUF
  
  • Urien Pascal
  , 2020, pp.1-2. (10.1109/CCNC46108.2020.9045502)
  DOI : 10.1109/CCNC46108.2020.9045502
• High Security Bare Metal Bluetooth Blockchain Payment Terminal For Trusted Ethereum Transaction
  
  • Urien Pascal
  , 2020, pp.1-2. (10.1109/CCNC46108.2020.9045146)
  DOI : 10.1109/CCNC46108.2020.9045146
• On the Effect of Aging on Digital Sensors
  
  • Anik Md Toufiq Hasan
  • Guilley Sylvain
  • Danger Jean-Luc
  • Karimi Naghmeh
  , 2020, pp.189-194. (10.1109/VLSID49098.2020.00050)
  DOI : 10.1109/VLSID49098.2020.00050
• Static Data-Flow Analysis of UML/SysML Functional Views for Signal and Image Processing Applications
  
  • Enrici Andrea
  • Apvrille Ludovic
  • Pacalet Renaud
  • Pham Minh Hiep
  , 2020, pp.101-126. (10.1007/978-3-030-37873-8_5)
  DOI : 10.1007/978-3-030-37873-8_5
• Nonparametric imputation by data depth
  
  • Mozharovskyi Pavlo
  • Josse Julie
  • Husson François
  Journal of the American Statistical Association, Taylor & Francis, 2020, 115 (529), pp.241-253. The presented methodology for single imputation of missing values borrows the idea from data depth --- a measure of centrality defined for an arbitrary point of the space with respect to a probability distribution or a data cloud. This consists in iterative maximization of the depth of each observation with missing values, and can be employed with any properly defined statistical depth function. On each single iteration, imputation is narrowed down to optimization of quadratic, linear, or quasiconcave function being solved analytically, by linear programming, or the Nelder-Mead method, respectively. Being able to grasp the underlying data topology, the procedure is distribution free, allows to impute close to the data, preserves prediction possibilities different to local imputation methods (k-nearest neighbors, random forest), and has attractive robustness and asymptotic properties under elliptical symmetry. It is shown that its particular case --- when using Mahalanobis depth --- has direct connection to well known treatments for multivariate normal model, such as iterated regression or regularized PCA. The methodology is extended to the multiple imputation for data stemming from an elliptically symmetric distribution. Simulation and real data studies positively contrast the procedure with existing popular alternatives. The method has been implemented as an R-package. (10.1080/01621459.2018.1543123)
  DOI : 10.1080/01621459.2018.1543123
• Time Series Source Separation with Slow Flows
  
  • Pineau Edouard
  • Razakarivony Sébastien
  • Bonald Thomas
  , 2020. In this paper, we show that slow feature analysis (SFA), a common time series decomposition method, naturally fits into the flow-based models (FBM) framework, a type of invertible neural latent variable models. Building upon recent advances on blind source separation, we show that such a fit makes the time series decomposition identifiable.
• DNN Based Beam Selection in mmW Heterogeneous Networks
  
  • Jagyasi Deepa
  • Coupechoux Marceau
  , 2020. We consider a heterogeneous cellular network wherein multiple small cell millimeter wave (mmW) base stations (BSs) coexist with legacy sub-6GHz macro BSs. In the mmW band, small cells use multiple narrow beams to ensure sufficient coverage and User Equipments (UEs) have to select the best small cell and the best beam in order to access the network. This process usually based on exhaustive search may introduce unacceptable latency. In order to address this issue, we rely on the sub-6GHz macro BS support and propose a deep neural network (DNN) architecture that utilizes basic components from the Channel State Information (CSI) of sub-6GHz network as input features. The output of the DNN is the mmW BS and beam selection that can provide the best communication performance. In the set of features, we avoid using the UE location, which may not be readily available for every device. We formulate a mmW BS selection and beam selection problem as a classification and regression problem respectively and propose a joint solution using a branched neural network. The numerical comparison with the conventional exhaustive search results shows that the proposed design demonstrate better performance than exhaustive search in terms of la-tency with at least 85% accuracy.
• Change Detection Based on the Coefficient of Variation in SAR Time-Series of Urban Areas
  
  • Colin Koeniguer Elise
  • Nicolas Jean-Marie
  Remote Sensing, MDPI, 2020, 12 (13), pp.2089. This paper discusses change detection in SAR time-series. First, several statistical properties of the coefficient of variation highlight its pertinence for change detection. Subsequently, several criteria are proposed. The coefficient of variation is suggested to detect any kind of change. Furthermore, several criteria that are based on ratios of coefficients of variations are proposed to detect long events, such as construction test sites, or point-event, such as vehicles. These detection methods are first evaluated on theoretical statistical simulations to determine the scenarios where they can deliver the best results. The simulations demonstrate the greater sensitivity of the coefficient of variation to speckle mixtures, as in the case of agricultural plots. Conversely, they also demonstrate the greater specificity of the other criteria for the cases addressed: very short event or longer-term changes. Subsequently, detection performance is assessed on real data for different types of scenes and sensors (Sentinel-1, UAVSAR). In particular, a quantitative evaluation is performed with a comparison of our solutions with baseline methods. The proposed criteria achieve the best performance, with reduced computational complexity. On Sentinel-1 images containing mainly construction test sites, our best criterion reaches a probability of change detection of 90% for a false alarm rate that is equal to 5%. On UAVSAR images containing boats, the criteria proposed for short events achieve a probability of detection equal to 90% of all pixels belonging to the boats, for a false alarm rate that is equal to 2%. (10.3390/rs12132089)
  DOI : 10.3390/rs12132089
• D’un simple dessin de Léonard de Vinci aux "formes premières
  
  • Crettez Jean-Pierre
  ISTE Openscience, ISTE Ltd, 2020, 4 (4). (10.21494/ISTE.OP.2020.0557)
  DOI : 10.21494/ISTE.OP.2020.0557
• Estimated whole-brain and lobe-specific radiofrequency electromagnetic fields doses and brain volumes in preadolescents
  
  • Cabré-Riera Alba
  • El Marroun Hanan
  • Muetzel Ryan
  • van Wel Luuk
  • Liorni Ilaria
  • Thielens Arno
  • Birks Laura
  • Pierotti Livia
  • Huss Anke
  • Joseph Wout
  • Wiart Joe
  • Capstick Myles
  • Hillegers Manon
  • Vermeulen Roel
  • Cardis Elisabeth
  • Vrijheid Martine
  • White Tonya
  • Röösli Martin
  • Tiemeier Henning
  • Guxens Mònica
  Environment International, Elsevier, 2020. Objective: To assess the association between estimated whole-brain and lobe-specific radiofrequency electromagnetic fields (RF-EMF) doses, using an improved integrated RF-EMF exposure model, and brain volumes in preadolescents at 9-12 years old. Methods: Cross-sectional analysis in preadolescents aged 9-12 years from the Generation R Study, a population-based birth cohort set up in Rotterdam, The Netherlands (n = 2592). An integrated exposure model was used to estimate whole-brain and lobe-specific RF-EMF doses (mJ/kg/day) from different RF-EMF sources including mobile and Digital Enhanced Cordless Telecommunications (DECT) phone calls, other mobile phone uses than calling, tablet use, laptop use, and far-field sources. Whole-brain and lobe-specific RF-EMF doses were estimated for all RF-EMF sources together (i.e. overall) and for three groups of RF-EMF sources that lead to a different pattern of RF-EMF exposure. Information on brain volumes was extracted from magnetic resonance imaging scans. Results: Estimated overall whole-brain RF-EMF dose was 84.3 mJ/kg/day. The highest overall lobe-specific dose was estimated in the temporal lobe (307.1 mJ/kg/day). Whole-brain and lobe-specific RF-EMF doses from all RF-EMF sources together, from mobile and DECT phone calls, and from far-field sources were not associated with global, cortical, or subcortical brain volumes. However, a higher whole-brain RF-EMF dose from mobile phone use for internet browsing, e-mailing, and text messaging, tablet use, and laptop use while wirelessly connected to the internet was associated with a smaller caudate volume. Conclusions: Our results suggest that estimated whole-brain and lobe-specific RF-EMF doses were not related to brain volumes in preadolescents at 9-12 years old. Screen activities with mobile communication devices while https://doi. T wirelessly connected to the internet lead to low RF-EMF dose to the brain and our observed association may thus rather reflect effects of social or individual factors related to these specific uses of mobile communication devices. However, we cannot discard residual confounding, chance finding, or reverse causality. Further studies on mobile communication devices and their potential negative associations with brain development are warranted, regardless whether associations are due to RF-EMF exposure or to other factors related to their use. (10.1016/j.envint.2020.105808)
  DOI : 10.1016/j.envint.2020.105808
• High resolution face age editing
  
  • Yao Xu
  • Newson Alasdair
  • Puy Gilles
  • Gousseau Yann
  • Hellier Pierre
  , 2020.
• Progressive Fusion for Unsupervised Binocular Depth Estimation using Cycled Networks
  
  • Pilzer Andrea
  • Lathuilière Stéphane
  • Xu Dan
  • Puscas Mihai Marian
  • Ricci Elisa
  • Sebe Nicu
  IEEE Transactions on Pattern Analysis and Machine Intelligence, Institute of Electrical and Electronics Engineers, 2020, 42 (10), pp.2380-2395. Recent deep monocular depth estimation approaches based on supervised regression have achieved remarkable performance. However, they require costly ground truth annotations during training. To cope with this issue, in this paper we present a novel unsupervised deep learning approach for predicting depth maps. We introduce a new network architecture, named Progressive Fusion Network (PFN), that is specifically designed for binocular stereo depth estimation. This network is based on a multi-scale refinement strategy that combines the information provided by both stereo views. In addition, we propose to stack twice this network in order to form a cycle. This cycle approach can be interpreted as a form of data-augmentation since, at training time, the network learns both from the training set images (in the forward half-cycle) but also from the synthesized images (in the backward half-cycle). The architecture is jointly trained with adversarial learning. Extensive experiments on the publicly available datasets KITTI, Cityscapes and ApolloScape demonstrate the effectiveness of the proposed model which is competitive with other unsupervised deep learning methods for depth prediction. (10.1109/TPAMI.2019.2942928)
  DOI : 10.1109/TPAMI.2019.2942928
• DYNAMIC-TDD INTERFERENCE TRACTABILITY APPROACHES AND PERFORMANCE ANALYSIS IN MACRO-CELL AND SMALL-CELL DEPLOYMENTS
  
  • Rachad J
  • Nasri R.
  • Decreusefond Laurent
  Annals of Telecommunications - annales des télécommunications, Springer, 2020. Meeting the continued growth in data traffic volume, Dynamic Time Division Duplex (D-TDD) has been introduced as a solution to deal with the uplink (UL) and downlink (DL) traffic asymmetry, mainly observed for dense heterogeneous network deployments, since it is based on instantaneous traffic estimation and provide more flexibility in resource assignment. However, the use of this feature requires new interference mitigation schemes capable to handle two additional types of interference between cells in opposite transmission direction: DL to UL and UL to DL interference. The aim of this work is to provide a complete analytical approach to model inter-cell interference in macro-cell and dense small-cell networks. We derive the explicit expressions of Interference to Signal Ratio (ISR) at each position of the network , in both DL and UL, to quantify the impact of each type of interference on the system performance. Also, we provide the explicit expressions of the coverage probability as functions of different system parameters by covering different scenarios. Finally, through system level simulations, we analyze the feasibility of D-TDD implementation in both deployments and we compare its performance to the static-TDD (S-TDD) configuration. (10.1007/s12243-020-00781-4)
  DOI : 10.1007/s12243-020-00781-4
• Spectral Mesh Simplification
  
  • Lescoat Thibault
  • Liu Hsueh-Ti Derek -
  • Thiery Jean-Marc
  • Jacobson Alec
  • Boubekeur Tamy
  • Ovsjanikov Maks
  Computer Graphics Forum, Wiley, 2020. The spectrum of the Laplace-Beltrami operator is instrumental for a number of geometric modeling applications, from processing to analysis. Recently, multiple methods were developed to retrieve an approximation of a shape that preserves its eigenvectors as much as possible, but these techniques output a subset of input points with no connectivity, which limits their potential applications. Furthermore, the obtained Laplacian results from an optimization procedure, implying its storage alongside the selected points. Focusing on keeping a mesh instead of an operator would allow to retrieve the latter using the standard cotangent formulation, enabling easier processing afterwards. Instead, we propose to simplify the input mesh using a spectrum-preserving mesh decimation scheme, so that the Laplacian computed on the simplified mesh is spectrally close to the one of the input mesh. We illustrate the benefit of our approach for quickly approximating spectral distances and functional maps on low resolution proxies of potentially high resolution input meshes.
• Private free-space communications based on chaos synchronization of mid-infrared quantum cascade laser light
  
  • Spitz O
  • Herdt A
  • Wu J
  • Wong C.-W
  • Elsässer W
  • Grillot F
  , 2020. Free Space Optics (FSO) is a growing technology offering higher bandwidth with fast and cost-effective deployment compared to fiber technology. This work demonstrates private free-space communication with quantum cascade lasers (QCLs). The secret message is encoded into a chaotic waveform so that the information is hard for an eavesdropper to extract [1]. Chaos-based transmissions in FSO are fundamentally restricted by atmospheric phenomena (e.g., turbulence, fog or scattering). Thus, the operating wavelength is a key parameter that has to be chosen wisely to reduce the impact of the environmental parameters. In this context, QCLs are relevant semiconductor lasers because their optical wavelength lies within mid-infrared domains where the atmosphere is highly transparent [2]. The simplest way to generate a chaotic optical carrier from a QCL is to feed back part of its emitted light into the device after a certain time delay [3], beyond which chaos synchronization between the drive and the response QCLs occurs.
• On generalized hyper-bent functions
  
  • Mesnager Sihem
  Cryptography and Communications-Discrete Structures, Boolean Functions, and Sequences (CCDS), 2020.
• New characterizations and construction methods of bent and hyper-bent Boolean functions
  
  • Mesnager Sihem
  • Mandal B.
  • Tang C.
  Discrete Mathematics, Elsevier, 2020.