Publications

2024

• Revisiting Hierarchical Text Classification: Inference and Metrics
  
  • Plaud Roman
  • Labeau Matthieu
  • Saillenfest Antoine
  • Bonald Thomas
  , 2024, Proceedings of the 28th Conference on Computational Natural Language Learning, pp.231-242. Hierarchical text classification (HTC) is the task of assigning labels to a text within a structured space organized as a hierarchy. Recent works treat HTC as a conventional multilabel classification problem, therefore evaluating it as such. We instead propose to evaluate models based on specifically designed hierarchical metrics and we demonstrate the intricacy of metric choice and prediction inference method. We introduce a new challenging dataset and we evaluate fairly, recent sophisticated models, comparing them with a range of simple but strong baselines, including a new theoretically motivated loss. Finally, we show that those baselines are very often competitive with the latest models. This highlights the importance of carefully considering the evaluation methodology when proposing new methods for HTC. (10.18653/v1/2024.conll-1.18)
  DOI : 10.18653/v1/2024.conll-1.18
• IEEE Circuits and Systems Society Tour de France 2024: Celebrating 75 Years of Innovation
  
  • Rivet François
  • Desgreys Patricia
  • Deltimple Nathalie
  • Frappé Antoine
  • Bourdel Sylvain
  • Latorre Laurent
  IEEE Circuits and Systems Magazine -New Series-, Institute of Electrical and Electronics Engineers, 2024, 24 (4), pp.25-34. The IEEE Circuits and Systems (CAS) Society France Chapter is excited to organize the “Tour de France” of Circuits and Systems in 2024, a celebratory event marking the 75th anniversary of the society. Inspired by the successful set of CASS Tours being organized in Region 9 since 2022, this event is a tribute to the sustained excellence and pioneering contributions of the IEEE CAS Society in the field of microelectronics. Aimed at strengthening the future of French and European technological sovereignty, the tour provides a unique opportunity for students, educators, and industry professionals to explore the latest advances and network with leading experts in the field. (10.1109/MCAS.2024.3410911)
  DOI : 10.1109/MCAS.2024.3410911
• Scalable and Reliable Decentralized Computing : From Asset Transfer to Atomic Snapshot
  
  • Araújo João Paulo Bezerra De
  , 2024. In recent years, the demand for decentralized data sharing has surged, as seen in applications like filesharing and cryptocurrency. While blockchain has been a popular solution for implementing decentralized applications, its high implementation costs due to the need for consensus on each new block pose significant challenges, especially in dynamic, decentralized settings. To address these issues, we explore solutions for abstractions that do not require consensus, such as reliable broadcast, atomic snapshot, and asset transfer systems.Focusing on asset-transfer systems, a key challenge is preventing double-spending, which traditionally imposes strong trust and synchrony assumptions. we take a nonorthodox approach to the double-spending problem that might suit better realistic environments in which these systemsare to be deployed. We consider the decentralized trust setting, where each user may independently choose who to trust by forming their local quorums. In this setting, we define k-Spending Asset Transfer, a relaxed version of asset transfer which bounds the number of times a participant may spend an asset it received. We establish a precise relationship between the decentralized trust assumptions and k, the optimal spending number of the system.For Byzantine reliable broadcast, a critical primitive in distributed systems, we tackle the scalability issue of quadratic message exchanges by proposing a probabilistic solution.This approach employs a small set of dynamically selected witnesses using a novel locality-preserving hash function, maintaining low latency and small communication complexity while tolerating slow adaptive adversaries.Lastly, we introduce a fast atomic-snapshot protocol for asynchronous message-passing systems and address issues in conventional time metrics for asynchronous long-lived implementations. Our new unified time-complexity analysis captures operation latency in various conditions, demonstrating significant latency improvements of our solution over state-of-the-art protocols, including optimal latency in fault-free runs and constant amortized latency.
• Reconfidencing LLM Uncertainty from the Grouping Loss Perspective
  
  • Chen Lihu
  • Perez-Lebel Alexandre
  • Suchanek Fabian
  • Varoquaux Gaël
  , 2024. <div><p>Large Language Models (LLMs), such as GPT and LLaMA, are susceptible to generating hallucinated answers in a confident tone. While previous efforts to elicit and calibrate uncertainty have shown some success, they often overlook biases towards certain groups, such as specific nationalities.</p><p>Existing calibration methods typically focus on average performance, failing to address this disparity. In our study, we demonstrate that the concept of grouping loss is an effective metric for understanding and correcting the heterogeneity in confidence levels. We introduce a novel evaluation dataset, derived from a knowledge base, specifically designed to assess the confidence scores of LLM responses across different groups. Our experimental results highlight significant variations in confidence, which are accurately captured by grouping loss. To tackle this issue, we propose a new method to calibrate the confidence scores of LLMs by considering different groups, a process we term reconfidencing. Our findings indicate that this approach effectively mitigates biases against minority groups, contributing to the development of fairer LLMs. The code is available at https: //github.com/tigerchen52/ reconfidencing_llms</p></div> (10.48550/arXiv.2402.04957)
  DOI : 10.48550/arXiv.2402.04957
• Statistical learning for multivariate and functional extremes
  
  • Huet Nathan
  , 2024. In a world where climate change is causing more and more extreme weather events of increasingmagnitude, this thesis explores the modeling of extreme events through statistical methods enhanced by statistical learning. It is divided into two main parts. First, functional extremes are studied, that is, the extremes of data explicitly dependent on a continuous variable such as time. We work in a separable Hilbert space, with a focus on the space L2[0; 1]. Results on regular variation, a fundamental hypothesis in extreme value theory, are developed, along with characterizations and non-trivial examples. Additionally, a dimensionality reduction method tailored to functional extreme data is proposed, with probabilistic and statistical guarantees. In the second part, we develop a probabilistic framework for regression in regions where the input variable is extreme, in contrast to classic approaches that focus on regions where the output variable is extreme. Results on risks and regression functions in extreme regions, as well as an adapted algorithm, are established. This algorithm is compared to classical methods and applied to the prediction of extreme sea levels in Brittany, where the goal is to reconstruct past extreme data to reduce uncertainties associated with certain estimates.
• Effect of narrowband filtering and pulse shaping for Distributed Acoustic Sensing
  
  • Chiabaut Anaïs
  • Arpison G.
  • Peigné A.
  • Gabet Renaud
  , 2024, pp.1-2. <div><p>Narrowband filtering to improve sensing range for Distributed Acoustic Sensing is investigated for two different pulse shapes and durations. Results from backscattered signal and phase noise measurements show that filtering is efficient providing that Modulation Instability remains the critical nonlinear phenomenon.</p></div> (10.1109/IPC60965.2024.10799571)
  DOI : 10.1109/IPC60965.2024.10799571
• A Contrastive Self-Supervised Learning scheme for beat tracking amenable to few-shot learning
  
  • Gagnere Antonin
  • Peeters Geoffroy
  • Essid Slim
  , 2024. In this paper, we propose a novel Self-Supervised-Learning scheme to train rhythm analysis systems and instantiate it for few-shot beat tracking. Taking inspiration from the Contrastive Predictive Coding paradigm, we propose to train a Log-Mel-Spectrogram Transformer encoder to contrast observations at times separated by hypothesized beat intervals from those that are not. We do this without the knowledge of ground-truth tempo or beat positions, as we rely on the local maxima of a Predominant Local Pulse function, considered as a proxy for Tatum positions, to define candidate anchors, candidate positives (located at a distance of a power of two from the anchor) and negatives (remaining time positions). We show that a model pre-trained using this approach on the unlabeled FMA, MTT and MTG-Jamendo datasets can successfully be fine-tuned in the few-shot regime, i.e. with just a few annotated examples to get a competitive beat-tracking performance.
• Using Pairwise Link Prediction and Graph Attention Networks for Music Structure Analysis
  
  • Buisson Morgan
  • Mcfee Brian
  • Essid Slim
  , 2024. The task of music structure analysis has been mostly addressed as a sequential problem, by relying on the internal homogeneity of musical sections or their repetitions. In this work, we instead regard it as a pairwise link prediction task. If for any pair of time instants in a track, one can successfully predict whether they belong to the same structural entity or not, then the underlying structure can be easily recovered. Building upon this assumption, we propose a method that first learns to classify pairwise links between time frames as belonging to the same section (or segment) or not. The resulting link features, along with node-specific information, are combined through a graph attention network. The latter is regularized with a graph partitioning training objective and outputs boundary locations between musical segments and section labels. The overall system is lightweight and performs competitively with previous methods. The evaluation is done on two standard datasets for music structure analysis and an ablation study is conducted in order to gain insight on the role played by its different components.
• Unlocking Regular Pulse Packages with Interband Cascade Lasers Grown on Si and GaSb substrates
  
  • Kim Hyunah
  • Poletti Thomas
  • Huang Heming
  • Díaz-Thomas Daniel Andrés
  • Fagot Maëva
  • Baranov A. N.
  • Cerutti Laurent
  • Grillot Frédéric
  , 2024, pp.1 - 2. <div><p>Optical feedback is studied on interband cascade lasers grown on Si and GaSb substrates. Experiments reveal the existence of regular pulse packages and suggest future nonlinear dynamic investigations for applications to free-space optics in the mid-infrared.</p></div> (10.1109/ipc60965.2024.10799831)
  DOI : 10.1109/ipc60965.2024.10799831
• Zero-Shot Structure Labeling with Audio And Language Model Embeddings
  
  • Buisson Morgan
  • Ick Christopher
  • Xi Tom
  • McFee Brian
  , 2024. Recent progress on audio-based music structure analysis has closely aligned with the appearance of new deep learning paradigms, notably for the extraction of robust spectro-temporal audio features and their sequential modeling. However, most recent methods resort to supervised learning, which requires careful annotation of audio music pieces. Such annotations may sometimes operate at different temporal scales from one dataset to another or comprise inconsistent variation markers across repetitions of identical segments. This work explores language models as an alternative to manual pre-processing of the section label space, thus facilitating training and predictions across different annotated corpora. We propose a joint audio-to-text embedding space in which latent representations of audio frames and their respective section labels are close. We take inspiration from recent works on cross-modal contrastive learning and demonstrate the plausibility of this paradigm in the context of music structure analysis.
• Malware detection through windows system call analysis
  
  • Hammi Badis
  • Hachem Joel
  • Rachini Ali
  • Khatoun Rida
  , 2024, pp.7. Detecting malware remains a significant challenge, as malware authors constantly develop new techniques to evade traditional signature-based and heuristic-based detection methods. This paper proposes a novel approach to malware detection that analyzes patterns in Windows system calls sequences to identify malicious behaviors. We use a voting classifier, a machine learning model that aggregates predictions from multiple individual models. It determines the final output based on the class that receives the highest likelihood or majority vote from the ensemble of models. We trained the model on large datasets of benign and malicious system call traces to detect anomalies indicative of malware. By focusing on system call behavior rather than static code characteristics, the approach is able to identify novel malware variants without requiring prior knowledge of their signatures. Experiments using a dataset of 42,797 API call sequences from malware samples and 1,079 sequences from benign software demonstrate that voting classifier can achieve high detectionrates while maintaining low false positive rates. This type of Machine Learning-based malware detection could be integrated into an Endpoint Detection and Response (EDR) tool to provide advanced, behavior-based malware detection capabilities.
• Malware detection through windows system call analysis
  
  • Hammi Badis
  • Hachem Joel
  • Rachini Ali
  • Khatoun Rida
  • Aissaoui Hassane
  , 2024, pp.1-7. Detecting malware remains a significant challenge, as malware authors constantly develop new techniques to evade traditional signature-based and heuristic-based detection methods. This paper proposes a novel approach to malware detection that analyzes patterns in Windows system calls sequences to identify malicious behaviors. We use a voting classifier, a machine learning model that aggregates predictions from multiple individual models. It determines the final output based on the class that receives the highest likelihood or majority vote from the ensemble of models. We trained the model on large datasets of benign and malicious system call traces to detect anomalies indicative of malware. By focusing on system call behavior rather than static code characteristics, the approach is able to identify novel malware variants without requiring prior knowledge of their signatures. Experiments using a dataset of 42,797 API call sequences from malware samples and 1,079 sequences from benign software demonstrate that voting classifier can achieve high detection rates while maintaining low false positive rates. This type of Machine Learning-based malware detection could be integrated into an Endpoint Detection and Response (EDR) tool to provide advanced, behavior-based malware detection capabilities. (10.1109/MobiSecServ63327.2024.10759991)
  DOI : 10.1109/MobiSecServ63327.2024.10759991
• Deep Learning Methods for Music Structure Analysis : Addressing Data Scarcity and Ambiguity
  
  • Buisson Morgan
  , 2024. Audio-based music structure analysis involves automatically identifying the different sections of an audio recording. Because various factors can influence human listeners’ perception of a music piece’s inner organization, musical structure is, by nature, subject to some ambiguity and subjectivity.In this dissertation, we focus on these characteristics of musical structure. We explore different research directions that explicitly incorporate them in the analysis process. The methods proposed in this thesis leverage different levels of supervision, either from a simple heuristic linked to the implicit timeline of music, from information contained in the input signal, or from data provided with structural annotations.First, we focus on the preliminary stage of music structure analysis, extracting audio features from the input signal. We opt for self-supervised representation learning approaches that can leverage large quantities of unlabelled music audio data. On the one hand, we propose a method directly inspired by the hierarchical aspect of music structure. On the other hand, we investigate another self-supervised strategy that uses the notion of repetition as an inductive bias to inform the frame sampling process.We then consider the scenario where we are given a set of music tracks for which human listeners have annotated the structure. In such a setting, we investigate how pre-trained audio representations may be adapted to such limited data corpora and highlight cases where the use of annotations can hurt segmentation performance.We finally introduce a fully-supervised approach to music structure analysis based on graph neural networks. The core of the method lies in extracting link features between time instants within the input track, which provide useful structural cues to predict segment boundaries and section labels.Our methods are evaluated on standard datasets for music structure analysis and systematically compared with recent state-of-the-art systems. Overall, the results suggest that explicitly targeting the issue of bias in structural annotations can help design more robust music structure analysis systems, which perform competitively against previous work even in limited training data situations.
• Neutron radiation tests and assessment of edge computing systems
  
  • Possamai Bastos Rodrigo
  • Laurini Luiz Henrique
  • Minelli de Carvalho Matheus
  • Naviner Lirida
  , 2024.
• Sequence Selection for Nonlinear Interference Mitigation: Current Approaches &amp; Open Challenges
  
  • Awwad Élie
  • Liu Jingtian
  • Hafermann Hartmut
  • Jaouën Yves
  , 2024, pp.1-6. In optical fiber transmission systems, nonlinear interference (NLI) significantly limits the achievable data rates for a reliable communication. In this talk, we review recent approaches to sequence selection as a method for minimizing NLI. By carefully selecting at the transmitter side symbol sequences that generate minimal NLI, it is possible to enhance system performance. We underline the various metrics used to assess the NLI generated by a symbol sequence in a transmission link, highlighting their theoretical bases and practical applications. Despite these recent advancements, numerous challenges remain unsolved, such as the complexity of predicting NLI accurately and on-the-fly at the transmitter, the computational burden of sequence selection, and the quest for novel schemes drawing upon the lessons learned from sequence selection. This talk discusses these open challenges and suggests potential directions for future research to address them. (10.1109/ACP/IPOC63121.2024.10809598)
  DOI : 10.1109/ACP/IPOC63121.2024.10809598
• IQ-Code for CDL and PDL Mitigation in Coupled Multi-Core Fiber
  
  • Abouseif Akram
  • Othman Ghaya Rekaya-Ben
  • Jaouën Yves
  • Darweesh Jamal
  • Klaimi Rami
  , 2024, pp.1-4. Low complex IQ-Code improves coupled MCF performance under PDL and CDL impairments, providing a 0.9 dB OSNR gain at FEC limit versus uncoded system. Employing IQ-code with core permutation, and ZF decoding achieves 99% complexity reduction over ML MIMO decoding with small OSNR penalty. (10.1109/ACP/IPOC63121.2024.10810082)
  DOI : 10.1109/ACP/IPOC63121.2024.10810082
• Effets environnementaux de la 5G (partie 2) : Applications envisagées et acteurs impliqués
  
  • Ciblat Philippe
  • Combaz Jacques
  • Coupechoux Marceau
  • Marquet Kevin
  • Orgerie Anne-Cécile
  1024 : Bulletin de la Société Informatique de France, Société Informatique de France, 2024 (24), pp.99-127. Cette série d'article a pour objectif de rassembler les connaissances actuelles liées aux effets environnementaux de la 5G. Elle s'articule autour de questions liées à ces effets et est organisée en trois parties : la première (publiée dans le bulletin 1024 de la SIF d'avril 2024) donnait les bases nécessaires à la compréhension de la technologie elle-même ; la deuxième, présentée ici, détaille les services rendus possibles par la 5G et les acteurs impliqués ; la dernière se penchera sur les effets environnementaux du déploiement de la 5G. Dans cette partie, nous décrivons les applications envisagées de la 5G et les stratégies des différents acteurs vis-à-vis de ces applications (consommateurs, États, opérateurs, fournisseurs de services, équipementiers, collectivités locales). Notre démarche est de décrire les applications permises par la 5G ainsi que les relations entre les différents acteurs de la 5G. Notre objectif est de nous servir des descriptions de cet objet socio-technologique pour ensuite évaluer sa pertinence écologique (qui sera développé dans la partie 3). Notre posture n'est pas de critiquer un acteur de la 5G en particulier, mais plutôt de pointer les intérêts divergents entre les différents acteurs. (10.48556/SIF.1024.24.99)
  DOI : 10.48556/SIF.1024.24.99
• Regenerative bootstrap for β-null recurrent Markov chains
  
  • Fernández Carlos
  Electronic Journal of Statistics, Shaker Heights, OH : Institute of Mathematical Statistics, 2024, 18, pp.4851-4881. Two regeneration-based bootstrap methods, namely, the Regeneration based-bootstrap [3, 19] and the Regenerative Block bootstrap [9] are shown to be valid for the problem of estimating the integral of a function with respect to the invariant measure in a β-null recurrent Markov chain with an accessible atom. An extension of the Central Limit Theorem for randomly indexed sequences is also presented.
• Knowledge Engineering and Knowledge Management : 24th International Conference, EKAW 2024, Amsterdam, The Netherlands, November 26–28, 2024, Proceedings
  
  • Alam Mehwish
  • Rospocher Marco
  • van Erp Marieke
  • Hollink Laura
  • Gesese Genet Asefa
  , 2024, 15370, pp.XV, 492. This book constitutes the refereed proceedings of the 24th International Conference on Knowledge Engineering and Knowledge Management, EKAW 2024, held in Amsterdam, The Netherlands, during November 26–28, 2024. The 28 full papers presented together were carefully reviewed and selected from 115 submissions. They focus on all aspects of knowledge in constructing systems and services for the semantic web, knowledge management, knowledge discovery, information integration, natural language processing, intelligent systems, e-business, e-health, humanities, cultural heritage, and beyond. (10.1007/978-3-031-77792-9)
  DOI : 10.1007/978-3-031-77792-9
• Model-Based Deep Learning for Music Information Research: Leveraging diverse knowledge sources to enhance explainability, controllability, and resource efficiency
  
  • Richard Gael
  • Lostanlen Vincent
  • Yang Yi-Hsuan
  • Müller Meinard
  IEEE Signal Processing Magazine, Institute of Electrical and Electronics Engineers, 2024, 41 (6). In this article, we investigate the notion of model-based deep learning in the realm of music information research (MIR). Loosely speaking, we refer to the term model-based deep learning for approaches that combine traditional knowledge-based methods with data-driven techniques, especially those based on deep learning, within a diff erentiable computing framework. In music, prior knowledge for instance related to sound production, music perception or music composition theory can be incorporated into the design of neural networks and associated loss functions. We outline three specifi c scenarios to illustrate the application of model-based deep learning in MIR, demonstrating the implementation of such concepts and their potential. (10.1109/MSP.2024.3415569)
  DOI : 10.1109/MSP.2024.3415569
• Towards More Robust NLP System Evaluation: Handling Missing Scores in Benchmarks
  
  • Himmi Anas
  • Irurozki Ekhine
  • Noiry Nathan
  • Clémençon Stéphan
  • Colombo Pierre
  , 2023, pp.11759-11785. The evaluation of natural language processing (NLP) systems is crucial for advancing the field, but current benchmarking approaches often assume that all systems have scores available for all tasks, which is not always practical. In reality, several factors such as the cost of running baseline, private systems, computational limitations, or incomplete data may prevent some systems from being evaluated on entire tasks. This paper formalize an existing problem in NLP research: benchmarking when some systems scores are missing on the task, and proposes a novel approach to address it. Our method utilizes a compatible partial ranking approach to impute missing data, which is then aggregated using the Borda count method. It includes two refinements designed specifically for scenarios where either task-level or instance-level scores are available. We also introduce an extended benchmark, which contains over 131 million scores, an order of magnitude larger than existing benchmarks. We validate our methods and demonstrate their effectiveness in addressing the challenge of missing system evaluation on an entire task. This work highlights the need for more comprehensive benchmarking approaches that can handle real-world scenarios where not all systems are evaluated on the entire task. (10.18653/v1/2024.findings-emnlp.688)
  DOI : 10.18653/v1/2024.findings-emnlp.688
• Metamaterial unipolar quantum optoelectronics for mid-infrared free-space optics
  
  • Bonazzi Thomas
  • Dely H.
  • Didier P.
  • Gacemi D.
  • Fix B.
  • Beck M.
  • Faist J.
  • Harouri A.
  • Sagnes I.
  • Grillot F.
  • Vasanelli A.
  • Sirtori C.
  APL Photonics, AIP Publishing LLC, 2024, 9 (11), pp.110801-1:110801-12. Free-space optical communications in the mid-infrared transparency windows (4–5 and 8–14 μm wavelength regions) is emerging as a viable solution for high bitrate data transmission. Unipolar quantum optoelectronics is the technology of choice for data communication in this wavelength region, thanks to the high frequency response of detectors and modulators. In this work, it is demonstrated that the performances of these devices can be substantially enhanced by embedding them into metamaterials. It is also shown that metamaterials have to be engineered differently in detectors than in modulators, as the role of light–matter interaction must be tuned adequately in the two devices. Metamaterial-enhanced performances allow the realization of data transmission with a record rate of 68 Gbit/s, while ensuring robustness and consistency, as it should be for real-world applications. These findings underscore the promising role of metamaterial-enhanced unipolar devices in advancing free-space optical communication systems. (10.1063/5.0225920)
  DOI : 10.1063/5.0225920
• Artificial intelligence and medical diagnosis: What are the ethical issues? Limits and fantasies, the researcher's point of view
  
  • Bloch Isabelle
  Journal d'imagerie diagnostique et interventionnelle, Elsevier, 2024, 8 (1), pp.45-48. While the applications of artificial intelligence in medical imaging are becoming increasingly numerous, the expected benefits are accompanied by challenges, and many questions remain open. This article, following a round table discussion at JFR 2023, summarizes the two main AI paradigms – based on knowledge representation and reasoning modeling for the former, and learning from large databases for the latter – and shows the advantages of combining them, in hybrid approaches. The question of bias, both statistical and cognitive, is addressed, as is that of explicability, central to the adoption of AI techniques. Links with ethical issues are also discussed. The conditions under which a system or algorithm was developed, the data used for training, the validation on independent bases, the conditions of use (type of image, type of problem or question to be solved…), biases and limits, are all parameters that need to be clearly defined. The role of the human beings remains essential, from the design of the methods to the interpretation of the results, framed in a multi-disciplinary approach. (10.1016/j.jidi.2024.09.004)
  DOI : 10.1016/j.jidi.2024.09.004
• End-to-End Service Guarantee for High-Speed Optical Networks
  
  • Soudais Guillaume
  , 2024. Driven by an ever-growing bandwidth and performance need, the IT network has grown such that OT and telecommunications networks are looking to exploit this infrastructure for their expansion. These three sectors have historically been separated due to different requirements, on latency, its variation and on reliability. To answer to time critical application needs, the Time Sensitive Network taskforce has developed new sets of protocols that are starting to be implemented in commercial products. Other groups have proposed novel architecture with time control to enable guaranteed performances between and inside edge datacenters. In my PhD I propose a solution to carry time critical application in legacy networks as it does not require to change the whole architecture. I show the benefits of its implementation in TSN networks for a future-proof solution with improved resource usage. To carry time critical traffic in legacy I propose to create a path by isolating and scheduling time critical traffic on a channel with guaranteed latency. With this construction, I build an algorithm to perform latency variation compensation enabling a constant latency transmission for time critical traffic. In a second time, propose a synchronization scheme and implemented a monitoring network primarily used here for latency monitoring, helping me gain insights on the distribution of latency that my protocol creates. Lastly, with an improved latency compensation algorithm, I demonstrate better jitter performances and study the turn-up time for our protocol enabling resource usage only when time-critical traffic is present. In my PhD I demonstrate, with an FPGA implementation and commercial product, latency variation reduction enabling OT and telecommunications network applications to run on legacy and TSN augmented network.
• Distributed Decoding Scheme for Uplink C-RAN System with Limited Backhaul Capacity
  
  • Chêne Thomas
  • Rekaya Ghaya
  • Damen Mohamed Oussama
  , 2024.