Publications

2023

• Near-field millimeter-wave radio-frequency exposure analysis
  
  • Jafari Seyedfaraz
  , 2023. This thesis aims to determine the absorbed power density (APD) considering the coupling and multiple reflections between the antenna and the human body, which poses challenges in assessing APD due to their close proximity.The first part of the thesis explores the concept of measuring APD inside a skin tissue phantom, specifically focusing on its application in 5G technologies.However, measuring APD inside the skin tissue phantom is limited due to the shallow penetration depth of fields at millimeter and quasi-millimeter waves. To overcome this limitation, a reconstruction technique is employed, utilizing the backward plane-wave spectrum(PWS) method. The electric field is sampled at a specific distance within the phantom, enabling the determination of APD at the human skin surface.In the second part, a non-invasive approach based on the dyadic Green's function (DGF) is proposed for APD assessment. This method takes into account the coupling between the human skin model and the device under test (DUT). The entire space is dividedinto two half-spaces : the upper half-space (z > 0) is filled with air, where the antenna is positioned, and the lower half-space is filled with an equivalent human skin liquid or solid. The electric field integral equation (EFIE), based on spatial DGFs, is solved using the method of moments (MoM) to reconstruct the equivalent currents. The electric field is sampled on the surface of a hemisphere surrounding the antenna, and the APD is evaluated based on the reconstructed equivalent currents beneath the air-phantom interface.In addition to the proposed techniques, the thesis investigates the measurement requirements for both approaches, including E-field measurement uncertainty, sampling angular resolution, and the required size of the phantom.The findings demonstrate that the proposed techniques present a novel methodology for assessing APD, taking into consideration the coupling between the human body and the antenna, particularly in the context of exposure to handheld devices operating above 6GHz.
• Estimation de la permittivité et conductivité de différents matériaux entre 2 GHz et 260 GHz
  
  • Aliouane Mohamed
  • Conrat Jean-Marc
  • Cousin Jean-Christophe
  • Begaud Xavier
  , 2023.
• Harris Recurrent Markov Chains and Nonlinear Monotone Cointegrated Models
  
  • Bertail Patrice
  • Durot Cécile
  • Fernández Carlos
  , 2023. In this paper, we study a nonlinear cointegration-type model of the form Z t = f 0 (X t)+W t where f 0 is a monotone function and X t is a Harris recurrent Markov chain. We use a nonparametric Least Square Estimator to locally estimate f 0 , and under mild conditions, we show its strong consistency and obtain its rate of convergence. New results (of the Glivenko-Cantelli type) for localized null recurrent Markov chains are also proved.
• Multi-target tracking and novel variational approaches for high-dimensional sequential data : an application to object counting in videos
  
  • Chagneux Mathis
  , 2023. This thesis studies novel approaches for prediction in high-dimensional sequential data, with a particular focus on scalability and weakly annotated settings. Motivated by the specific task of macrolitter counting in videos, we first derive a robust solution based on the multi-target tracking methodology, which combines deep learning and classical recursive Bayesian estimation.Then, we focus on the recent extensions of variational inference methods to the sequential setting, motivated by recent formulations of multi-target problems as statistical estimates in high-dimensional latent data models. Here, we study theoretical and computational aspects of the so-called backward factorization as a promising parameterization for variational approximations of the smoothing distributions in general state-space models. In particular, under such backward variational approximations, we derive a theoretical bound for the approximation error when considering expectations of additive state functionals, and develop an efficient algorithm for recursive latent estimation and online learning of the variational parameters. These contributions strengthen the relevance of variational approaches as alternatives to Monte Carlo methods in sequential settings, and pave the way to their adoption as generic solutions for unsupervised prediction in high-dimensional temporal data, such as for video object counting.
• Experimental evaluation of Rideability Index as a Handling quality indicator
  
  • Ronné Jules
  • Dubuis Laura
  • Robert Thomas
  , 2023. Understanding and mastering handling quality is a critical concern for bicycle designers, as it directly impacts safety, comfort, and performance. However, this aspect has received limited attention to date. Existing literature offers experimental handling quality indicators based on bicycle kinematics, but their validity has yet to be established. This study aims to assess the predictive power of these indicators using experimental data derived from subjective assessments of handling quality. These data, obtained from a protocol involving 20 participants and 2 bicycles, enabled testing 39 experimental indicators. The results indicate that certain vehicle kinematic quantities are indeed correlated with the perception of handling quality but with low predictive power. Indicators based on handlebar movement are the most effective in explaining the sensation of handling quality. These indicators perform particularly well at low speeds, where physical and cognitive workload are associated with the quantity of control actions on the handlebars.
• Computing the cohomology of constructible \'etale sheaves on curves
  
  • Levrat Christophe
  , 2023. We present an explicit expression of the cohomology complex of a constructible sheaf of abelian groups on the small \'etale site of an irreducible curve over an algebraically closed field, when the torsion of the sheaf is invertible in the field. This expression only involves finite groups, and is functorial in both the curve and the sheaf. In particular, we explain how to compute the Galois action on this complex. We also present an algorithm which computes it and study its complexity.
• Equalization in Optical Fiber Communication Using Model-based Neural Networks
  
  • Abu-Romoh Mohanad
  , 2023. Meeting the increasing demand for highspeed data transmission requires effective solutions for mitigating nonlinearity in optical communication systems. Traditional methods like Digital Backpropagation (DBP) face significant computational challenges,rendering them impractical for real-world systems. Neural network models have emerged as a promising approach to address this issue. Two primary approaches exist for designing neural networks: model-agnostic and model-based methods. Modelagnostictechniques offer flexibility in terms of size and hyperparameters and can be placed at various positions in the receiver’s Digital Signal Processing (DSP) chain. However, they demand substantial size and extensive training data in order to operate effectively.In contrast, model-based approaches employ neuralnetworks guided by the physical model of signal propagation. These approaches tend to be more compactbut require careful initialization for proper generalization. One prominent model-based technique isLearned Digital Backpropagation (LDBP), which optimizes DBP parameters using neural networks. LDBPoffers the promise of improved performance or reduced complexity compared to DBP. This study primarilyfocuses on LDBP, introducing simplifications through ”parameter sharing” to reduce trainable parameters.Additionally, we propose repurposing the legacy Dispersion-Managed (DM) systems, by incorporatinghigher-order modulation formats such as 16- QAM and 64-QAM, to enhance data rates within these systems. A comprehensive analysis of the performance and complexity demonstrates that the proposed algorithms outperform linear equalization andDBP in various transmission systems.
• Security Assessment Against Side-Channel Attacks : Insights from an Information-Theoretic Perspective
  
  • Liu Yi
  , 2023. In today's world, the widespread use of cryptographic devices highlights the need for their secure operation. Unintended leakages, like time, power, and electromagnetic emissions, can allow attackers to deduce secret keys via side-channel attacks (SCAs). Evaluating the security of cryptographic devices against SCAs is important for both the industrial and academic sectors, and information-theoretic metrics turn out to be effective tools. “Masking” stands out as a key countermeasure, with ongoing discussions on its optimization and the security of its implementations. In light of this context, the central aims of this thesis are to quantify side-channel leakage, appraise the security of cryptographic devices against SCAs (both unprotected and masked), and to explore methodologies for formulating more potent masking codes. For masking code construction, we establish linear programming bounds for the kissing number of q-ary linear codes, guided by recent findings on optimized code-based masking performance related to the dual code's kissing number. In addition, we demonstrate the connection between code-based masking efficacy and the whole weight enumeration of the dual of the masking code. The lexicographical order based on weight distribution prefixes is proposed for selecting ideal masking codes. Regarding side-channel leakage evaluation, we introduce a novel information metric, called conditional Sibson's alpha-information. This metric has an explicit expression and possesses several beneficial properties. Utilizing this metric, we delve into the sidechannel leakage of unprotected devices. Additionally, we use Fano's mutual information to evaluate the sidechannel leakage of code-based masked implementations under probing model. Lastly, when considering the security assessment of masked implementations, we utilize the alphainformation measure to appraise the security of both arithmetic and Boolean masking implementations. We derive universal bounds on the probability of success of any type of side-channel attack. These also provide lower bounds on the minimum number of queries required to achieve a given success rate.
• Curves are algebraic $K(\pi,1)$: theoretical and practical aspects
  
  • Levrat Christophe
  , 2023. We prove that any connected curve with a rational point $x$ over a field $k$ is an algebraic $K(\pi,1)$, as soon as all of its irreducible components have nonzero genus. This means that the cohomology of any locally constant constructible \'etale sheaf of $\mathbb{Z}/n\mathbb{Z}$-modules, with $n$ invertible in $k$, is canonically isomorphic to the cohomology of its corresponding $\pi_1(X,\bar x)$-module, where $\bar x$ is a geometric point of $X$ above $x$. When $k$ is algebraically closed, we explicitly describe finite quotients of $\pi_1(X,\bar x)$ that allow to compute the cohomology groups of the sheaf, as well as the cup products $H^1\times H^1\to H^2$.
• A Tool for Investigating Cyber-Physical Systems via SystemC AMS Virtual Prototypes Derived from SysML Models
  
  • Genius Daniela
  • Apvrille Ludovic
  , 2023, pp.1-6. This contribution introduces an open-source soft-ware tool, called TTool-AMS, designed to streamline the design process of cyber-physical systems. It achieves this by enabling the direct generation of SystemC AMS virtual prototypes from Systems Modeling Language(SysML) models. This tool acknowledges the diversity of Models of Computation (MoC) in the design process, accommodating three types of SystemC AMS MoC and their respective conversion interfaces. It aims to simplify early-stage integration and enhance exploration of the interactions between analog and digital components in cyber-physical systems.
• A formal framework to design and prove trustworthy memory controllers
  
  • Lisboa Malaquias Felipe
  • Asavoae Mihail
  • Brandner Florian
  Real-Time Systems, Springer Verlag, 2023, 59 (4), pp.664-704. Abstract In order to prove conformance to memory standards and bound memory access latency, recently proposed real-time DRAM controllers rely on paper and pencil proofs, which can be troubling: they are difficult to read and review, they are often shown only partially and/or rely on abstractions for the sake of conciseness, and they can easily diverge from the controller implementation, as no formal link is established between both. We propose a new framework written in Coq, in which we model a DRAM controller and its expected behaviour as a formal specification. The trustworthiness in our solution is two-fold: (1) proofs that are typically done on paper and pencil are now done in Coq and thus certified by its kernel, and (2) the reviewer’s job develops into making sure that the formal specification matches the standards—instead of performing a thorough check of the mathematical formalism. Our framework provides a generic DRAM model capturing a set of controller properties as proof obligations, which all implementations must comply with. We focus on properties related to the assertiveness that timing constraints are respected, every incoming request is handled in bounded time, and the DRAM command protocol is respected. We refine our specification with two implementations based on widely-known arbitration policies— First-in First-Out (FIFO) and Time-Division Multiplexing (TDM). We extract proved code from our model and use it as a “trusted core” on a cycle-accurate DRAM simulator. (10.1007/s11241-023-09411-3)
  DOI : 10.1007/s11241-023-09411-3
• Audiocarnet for the presentation of the Statistical Wave Field Theory
  
  • Badeau Roland
  , 2023. Audiocarnet du CNRS pour la journée « Action Audio » du GdR ISIS le 16/11/23 à l'IRCAM, sur le thème « Traitement du signal pour la musique ».
• Free-Space Gigabit Data Transmission with a Directly Modulated Interband Cascade Laser Epitaxially Grown on Silicon
  
  • Zaminga Sara
  • Didier Pierre
  • Kim Hyunah
  • Díaz-Thomas D.
  • Baranov A.
  • Rodriguez J.
  • Tournié E.
  • Knötig H.
  • Spitz O.
  • Schwarz B.
  • Cerutti L.
  • Grillot Frédéric
  , 2023, pp.1-2. This work demonstrates mid-infrared data transmission using an epitaxial interband cascade laser on silicon emitting at 3.5 μm. Direct modulation at 5 Gbit/s with on-off keying is achieved at room temperature while digital signal processing is used to further improve the system performance. (10.1109/IPC57732.2023.10360737)
  DOI : 10.1109/IPC57732.2023.10360737
• Procédé de calcul d’un pire temps de transmission, programme d’ordinateur et système informatique associés
  
  • Champenois Florient
  • Brandner Florian
  • Grandpierre Thierry
  , 2023.
• Ontology Matching using Textual Class Descriptions
  
  • Peng Yiwen
  • Alam Mehwish
  • Bonald Thomas
  , 2023. In this paper, we propose TEXTO, a TEXT-based Ontology matching system. This matcher leverages the rich semantic information of classes available in most ontologies by a combination of a pre-trained word embedding model and a pre-trained language model. Its performance is evaluated on the datasets of the OAEI Common Knowledge Graphs Track, augmented with the description of each class, and a new dataset based on the refreshed alignment of Schema.org and Wikidata. Our results demonstrate that TEXTO outperforms all state-of-art matchers in terms of precision, recall, and F1 score. In particular, we show that almost perfect class alignment can be achieved using textual content only, excluding any structural information like the graph of classes or the instances of each class.
• Reasoning About Dynamic Game Models Using Obstruction Logic (short paper)
  
  • Catta Davide
  • Leneutre Jean
  • Malvone Vadim
  , 2023, 3585. Games played within dynamic models have been explored in various domains, including cybersecurity and planning. Our paper introduces Obstruction Logic, a formalism designed for analyzing specific games featuring temporal objectives, which unfold within dynamic models. These games involve players whose actions can impact the underlying game model. We demonstrate how this logic can be employed to express significant properties within the realm of cybersecurity games, particularly those defined on attack graphs. An expanded version of our research has been published in ECAI 2023.
• Refinements for Open Automata
  
  • Ameur-Boulifa Rabéa
  • Corradi Quentin
  • Henrio Ludovic
  • Madelaine Eric
  , 2023, LNCS-14323, pp.11-29. Establishing equivalence and refinement relations between programs is an important mean for verifying their correctness. By establishing that the behaviours of a modified program simulate those of the source one, simulation relations formalise the desired relationship between a specification and an implementation, two equivalent implementations, or a program and its optimised implementation. This article discusses a notion of simulation between open automata, which are symbolic behavioural models for communicating systems. Open automata may have holes modelling elements of their context, and can be composed by instantiation of the holes. This allows for a compositional approach for verification of their behaviour. We define a simulation between open automata that may or may not have the same holes, and show under which conditions these refinements are preserved by composition of open automata. (10.1007/978-3-031-47115-5_2)
  DOI : 10.1007/978-3-031-47115-5_2
• Unexpected Attributed Subgraphs: a Mining Algorithm
  
  • Delarue Simon
  • Viard Tiphaine
  • Dessalles Jean-Louis
  , 2023, pp.171-178. Graphs are ubiquitous in real-world data, ranging from the study of social interactions to bioinformatics or the modelling of physical systems. These real-world graphs are typically sparse, possibly large and frequently contain additional information in the form of attributes, making them a complex object to understand. Graph summarization techniques can help facilitate the discovery of hidden patterns in underlying data by providing an interesting subset of the interactions and available attributes, which we broadly call a pattern. However, determining what is considered interesting in this context is not straightforward. We address this challenge by designing an interestingness measure based on the information-theoretic measure of Unexpectedness, linking the concepts of relevance and Kolmogorov complexity. We design a pattern mining algorithm to provide a summary of the initial data in the form of a set of unexpected patterns, that is, patterns for which there is a drop between their expected complexity and the observed complexity. Experimental results on five real-world datasets with state-of-the-art methods demonstrate that our method exhibits a small number of diversified patterns, providing a humanreadable summary of the initial attributed graph; we show that our summaries quantitatively outperforms attribute-only and interaction-only baselines as well as other pattern mining methods, reinforcing the need for methods dealing with attributed graphs. We visualize summaries extracted with our method, in order to qualitatively validate their readability. (10.1145/3625007.3627314)
  DOI : 10.1145/3625007.3627314
• Strong converses using typical changes of measures and asymptotic Markov Chains
  
  • Hamad Mustapha
  • Wigger Michèle
  • Sarkiss Mireille
  IEEE Transactions on Information Theory, Institute of Electrical and Electronics Engineers, 2023, pp.i-xxvi. The paper presents exponentially-strong converses for source-coding, channel coding, and hypothesis testing problems. More specifically, it presents alternative proofs for the well-known exponentially-strong converse bounds for almost lossless source-coding with side-information and for channel coding over a discrete memoryless channel (DMC). These alternative proofs are solely based on a change of measure argument on the sets of conditionally or jointly typical sequences that result in a correct decision, and on the analysis of these measures in the asymptotic regime of infinite blocklengths. The paper also presents a new exponentially-strong converse for the K-hop hypothesis testing against independence problem with certain Markov chains and a strong converse for the two-terminal Lround interactive compression problem with multiple distortion constraints that depend on both sources and both reconstructions. This latter problem includes as special cases the Wyner-Ziv problem, the interactive function computation problem, and the compression with lossy common reconstruction problem. These new strong converse proofs are derived using similar change of measure arguments as described above and by additionally proving that certain Markov chains involving auxiliary random variables hold in the asymptotic regime of infinite blocklengths. As shown in related publications, the same method also yields converse bounds under expected resource constraints. (10.1109/TIT.2023.3327314)
  DOI : 10.1109/TIT.2023.3327314
• Self-Similarity-Based and Novelty-based loss for music structure analysis
  
  • Peeters Geoffroy
  , 2023. Music Structure Analysis (MSA) is the task aiming at identifying musical segments that compose a music track and possibly label them based on their similarity. In this paper we propose a supervised approach for the task of music boundary detection. In our approach we simultaneously learn features and convolution kernels. For this we jointly optimize - a loss based on the Self-Similarity- Matrix (SSM) obtained with the learned features, denoted by SSM-loss, and - a loss based on the novelty score obtained applying the learned kernels to the estimated SSM, denoted by novelty-loss. We also demonstrate that relative feature learning, through self-attention, is beneficial for the task of MSA. Finally, we compare the performances of our approach to previously proposed approaches on the standard RWC-Pop, and various subsets of SALAMI.
• A Repetition-based Triplet Mining Approach for Music Segmentation
  
  • Buisson Morgan
  • Mcfee Brian
  • Essid Slim
  • Crayencour Helene-Camille
  , 2023. Contrastive learning has recently appeared as a well-suited method to find representations of music audio signals that are suitable for structural segmentation. However, most existing unsupervised training strategies omit the notion of repetition and therefore fail at encompassing this essential aspect of music structure. This work introduces a triplet mining method which explicitly considers repeating sequences occurring inside a music track by leveraging common audio descriptors. We study its impact on the learned representations through downstream music segmentation. Because musical repetitions can be of different natures, we give further insight on the role of the audio descriptors employed at the triplet mining stage as well as the trade-off existing between the quality of the triplets mined and the quantity of unlabelled data used for training. We observe that our method requires less non-annotated data while remaining competitive against other unsupervised methods trained on a larger corpus.
• PESTO: Pitch Estimation with Self-supervised Transposition-equivariant Objective
  
  • Riou Alain
  • Lattner Stefan
  • Hadjeres Gaëtan
  • Peeters Geoffroy
  , 2023. In this paper, we address the problem of pitch estimation using Self Supervised Learning (SSL). The SSL paradigm we use is equivariance to pitch transposition, which enables our model to accurately perform pitch estimation on monophonic audio after being trained only on a small unlabeled dataset. We use a lightweight ($<$ 30k parameters) Siamese neural network that takes as inputs two different pitch-shifted versions of the same audio represented by its Constant-Q Transform. To prevent the model from collapsing in an encoder-only setting, we propose a novel class-based transposition-equivariant objective which captures pitch information. Furthermore, we design the architecture of our network to be transposition-preserving by introducing learnable Toeplitz matrices. We evaluate our model for the two tasks of singing voice and musical instrument pitch estimation and show that our model is able to generalize across tasks and datasets while being lightweight, hence remaining compatible with low-resource devices and suitable for real-time applications. In particular, our results surpass self-supervised baselines and narrow the performance gap between self-supervised and supervised methods for pitch estimation. (10.48550/arXiv.2309.02265)
  DOI : 10.48550/arXiv.2309.02265
• Transfer Learning and Bias Correction with Pre-trained Audio Embeddings
  
  • Wang Changhong
  • Richard Gaël
  • Mcfee Brian
  , 2023. Deep neural network models have become the dominant approach to a large variety of tasks within music information retrieval (MIR). These models generally require large amounts of (annotated) training data to achieve high accuracy. Because not all applications in MIR have sufficient quantities of training data, it is becoming increasingly common to transfer models across domains. This approach allows representations derived for one task to be applied to another, and can result in high accuracy with less stringent training data requirements for the downstream task. However, the properties of pre-trained audio embeddings are not fully understood. Specifically, and unlike traditionally engineered features, the representations extracted from pre-trained deep networks may embed and propagate biases from the model's training regime. This work investigates the phenomenon of bias propagation in the context of pre-trained audio representations for the task of instrument recognition. We first demonstrate that three different pre-trained representations (VGGish, OpenL3, and YAMNet) exhibit comparable performance when constrained to a single dataset, but differ in their ability to generalize across datasets (OpenMIC and IRMAS). We then investigate dataset identity and genre distribution as potential sources of bias. Finally, we propose and evaluate post-processing countermeasures to mitigate the effects of bias, and improve generalization across datasets.
• Analysis of the singularity avoidance capability of Constant Modulus Algorithms in coherent optical fibre communication systems
  
  • Nwakamma Peter A.
  • Froc Gwillerm
  • Jaouën Yves
  • Ware Cédric
  , 2023. In dual-polarization coherent optical fibre transmission, adaptive blind equalizers based on constant modulus algorithm (CMA) have been proposed for polarization demultiplexing and inter-symbol interference compensation since it allows good convergence at low computational complexity. While CMA is known to be hampered by singularities induced by correlations between the equalizer’s output signals, it has been shown that their occurrences can be avoided by modifying the CMA scheme o penalize these correlations. However, the dependence of the singularity avoidance according to both the parameters of the CMAs and the channel has not been exhaustively studied. Hence, considering optical access networks with the objective to avoid CMA singularity in a wide range of polarization mode dispersion (PMD) situations, we study this dependence by computer simulations over deterministic realizations of the channel. We introduce a refined CMA scheme and exhibit proper sets of parameters for singularity-free operation up to 5 ps/√km PMD realization (22.36 ps DGD) in a 32 GBaud dual-polarization QPSK system. Solutions for covering extended PMD ranges are suggested also.
• Fault-Tolerant Four-Dimensional Constellation for Coherent Optical Transmission Systems
  
  • Liu Jingtian
  • Awwad Élie
  • Jaouën Yves
  , 2023. We propose a 4-dimensional 2-ary amplitude ring-switched modulation format with 64 symbols, which is denoted as 4D-2A-RS64 encoded over two polarization tributaries to improve the transmission performance over long-haul optical fibers in the presence of the non-linear Kerr effect. At a spectral efficiency of 6 bits per 4D, simulation results show that this format outperforms the polarization division multiplexed (PDM) 8QAM-star modulation as well as the 4D-2A-8PSK over links without inline dispersion management. We evaluate the performance for a WDM transmission of 11 × 90 Gbaud channels over a multi-span SSMF link. For an achievable information rate of 4.8bit/s/Hz, the maximum transmission distance is improved by 10.6% (400 km) and 4% (160 km) compared to PDM-8QAM-star and 4D-2A-8PSK respectively. The achieved gains are composed of a linear part and a non-linear part, respectively from the improved Euclidean-distance distribution and the constant power property of the 4D modulation. The geometric shaping of the proposed scheme is easy to implement and is robust to Mach-Zehnder modulator (MZM) imbalances and quantization errors stemming from the finite digital-to-analog converter (DAC) resolution. This robustness is compared to the one of other geometric-shaped non-linearity tolerant 4D schemes such as the 4D-2A-8PSK and the 4D-64PRS that can be both outperformed by our scheme in severe conditions.