Search

International audience; Cognition of user behavior can make future mobile networks more intelligent and flexible. Knowledge about users' habits can be used to personalize services and intelligently manage network resources. However, inferring this key information with a low-cost signaling implementation, and avoiding constant user interaction, is crucial for Mobile Network Operators (MNOs). With this motivation, this paper investigates the detection of the real-life mobile user environment using contextaware detection via multi-task learning (MTL). We propose models that are able to automatically detect up to eight distinct real-life user environments. We also improve the detection accuracy with the assistance of the mobility state profiling task. We associate both environment and mobility tasks because they correspond to the main attributes of user behavior and, additionally, both of them are correlated. Using MTL, the task of detecting environment corresponds to simultaneously answering the questions: "how and where mobile user consumes mobile services?". We build models using real-life radio data which is already available in network. This data has been massively gathered from multiple diversified situations of mobile users. Simulation results support our claim to detect several environment classes in network infrastructure with improved UED accuracy.

International audience; Resilience is defined as the ability of a network to resist, adapt to and quickly bounce back from disruptions, to continue keeping an acceptable level of service from users' perspective. To characterize and analyse the resilience of a 5G Radio Access Network, it is vital to be able to measure current and prospective resiliency levels using relevant metrics. In this work, we perform an analysis and a quantification of 5G-RAN resilience based on radio coverage indicator and one of the Quality of Service indicators, namely the service integrity. They are considered as the main performance indicators for network planners and operators. In this objective, we model network states using the performance indicators, Reference Signal Received Power and Received User Equipment Throughput. Then, we use them to build Markovian models and to define resilience metrics. By means of simulations, we evaluate the ability of our analytical model to accurately capture network resiliency levels. We discuss numerical results from multiple usage scenarios to show how the resilience metrics can be used to identify potential outages and to help network planners or operators deciding on which adequate adaptive resilient mechanisms to deploy.

We characterize trade-offs between the end-to-end communication delay and the energy in urban vehicular communications with infrastructure assistance. Our study exploits the self-similarity of the location of communication entities in cities by modeling them with the hyperfractal model which charaterize the distribution of mobile nodes and relay nodes by a fractal dimension d F and d r , both larger than the dimension of the embedded map. We compute theoretical bounds for the end-to-end communication hop count considering two different energy-minimizing goals: either total accumulated energy or maximum energy per node. Let δ > 1 the attenuation factor in the street, we prove that when we aim to a total energy cost of order n (1−δ)(1−α) the hop count for an end-to-end transmission is of order n 1−α/(d F −1) , with α < 1 is a tunable parameter. This proves that for both goals the energy decreases as we allow choosing routing paths of higher length. The asymptotic limit of the energy becomes significantly small when the number of nodes becomes asymptotically large. A lower bound on the network throughput capacity with constraints on path energy is also given. We show that our model fits real deployments where open data sets are available. The results are confirmed through simulations using different fractal dimensions in a Matlab simulator.

International audience; Consumer Internet of Things (IoT) applications are largely built through end-user programming in the form of event-action rules. Although, end-user tools help simplify the building of IoT applications to a large extent, there are still challenges in developing expressive applications in a simple yet correct fashion. In this context, we propose a formal development framework based on the Web of Things specification. An application is defined using a composition language which allows users to compose the basic event-action rules to express complex scenarios. It is transformed into a formal specification which serves as the input for formal analysis, where the application is checked for functional and quantitative properties at design time using model checking techniques. Once the application is validated, it can be deployed and the rules are executed following the composition language semantics. We have implemented these proposals in a tool built on top of Mozilla WebThings platform. The steps from design to deployment were validated on real-world applications.

International audience; A 5G Radio Access Network (5G-RAN) can be disturbed or shutdown due to a variety of failures, in spite of advanced optimization techniques and self-healing methods. Recently, operators started to take an interest in improving the resilience of communication networks with adaptive compensation techniques to mitigate outage-induced performance degradation. However, for the sake of effective and efficient resilience management, it is vital to be able to measure current and prospective resiliency levels of a given 5G-RAN using relevant and explicit metrics. Thus, the characterisation of resilience goes over a thorough analysis of 5G-RAN performance indicators followed by a rigorous quantification of current and future levels of resilience. In this work, we perform an analysis and a quantification of 5G-RAN resilience using a coverage indicator. It is known as a main performance indicator for network planners and operators, coverage is a necessary prerequisite to ensure a certain level of Quality of Service. For that, we model the network coverage using Continuous Time Markov Chains (CTMCs) where coverage status is characterized with multiple states defined with Reference Signal Received Power (RSRP) signal. The proposed Markov model is analytically studied allowing to perform quantitative analysis, predict coverage outage and provide resilience quantification. Using our model, we conduct numerical analysis of several usage scenarios and propose a resilience framework to show the usability of our proposed approach.

International audience; Network slicing is one of the building blocks in Zero Touch Networks. It mainly consists in a dynamic deployment of services in a substrate network. However, the Virtual Network Embedding (VNE) algorithms used generally follow a static mechanism, which results in sub-optimal embedding strategies and less robust decisions. Some reinforcement learning algorithms have been conceived for a dynamic decision, while being time-costly. In this paper, we propose a combination of deep Q-Network and a Monte Carlo (MC) approach. The idea is to learn, using DQN, a distribution of the placement solution, on which a MC-based search technique is applied. This improves the solution space exploration, and achieves a faster convergence of the placement decision, and thus a safer learning. The obtained results show that DQN with only 8 MC iterations achieves up to 44% improvement compared with a baseline First-Fit strategy, and up to 15% compared to a MC strategy.

International audience; In this paper, three different designs of test structures are explored in order to accurately characterize InP DHBTs up to 110 GHz. In particular, a new design, optimized for high frequency measurements while keeping high device density, has been proposed. De-embedding test structures are analyzed and InP DHBT RF figures of merit are extracted for the three designs. Extraction of the maximum oscillation frequency, fMAX, confirms the relevance of optimized test structures as well as good performances of the new design.

International audience; This work aims at inferring a set of regular expressions to parse a text file, like a system log. To this end, we propose a novel edit distance taking advantage of the pattern matching background. Edit distances are commonly used for fuzzy search and in bioinformatics, and compare two strings at the character level. By doing so, edit distances do not consider the nature of the data conveyed by the strings. To address this problem, we propose the following contributions. First, we propose to model strings at the pattern level using a dedicated data structure, called pattern automaton. Second, we design a novel edit distance, operating at the pattern level. Third, we derive a clustering algorithm optimized for this distance. Finally, we evaluate our proposal through experimental validation.

International audience; We address the issue of maximizing the number of connected devices in a Narrowband Internet of Things (NB-IoT) network using non-orthogonal multiple access (NOMA) in the downlink. We first propose an optimal joint sub-carrier and power allocation strategy assuming perfect channel state information (CSI) called Stratified Device Allocation (SDA), that maximizes the connectivity under data rate, power and bandwidth constraints. Then, we generalize the connectivity maximization problem to the case of partial CSI, where only the distancedependent path-loss component of the channel gain is available at the base station (BS). We introduce a novel framework called the Stochastic Connectivity Optimization (SCO) framework. In this framework, we propose a heuristic improvement to SDA namely SDA with Excess Power (SDA-EP) algorithm for operation under partial CSI. Furthermore, we derive a concave approximation (SCO-CA) algorithm of near-optimal performance to SCO given the same amount of CSI. Through computer simulations, we show that SDA-EP and SCO-CA outperform conventional NOMA and OMA schemes in the presence of partial CSI over a wide range of service scenarios.

We demonstrate the detection of speech signals and footsteps through a novel polarization-multiplexed coded differential-phase OTDR (optical time domain reflectometry) interrogation of standard telecom fibers in different configurations. We assess the sensitivity to audio signals by measuring the induced phase variation with respect to the sound pressure level of the perturbation, and manage to detect acoustic signals with high reliability. Further experiments are performed using an installed fiber cable showing the ability to track footsteps in an indoor scenario. These experiments pave the way for numerous applications of Distributed Acoustic Sensing (DAS) where a high sensitivity is required.

In this paper, we apply an multi-agent reinforcement learning (MARL) framework allowing the base station (BS) and the user equipments (UEs) to jointly learn a channel access policy and its signaling in a wireless multiple access scenario. In this framework, the BS and UEs are reinforcement learning (RL) agents that need to cooperate in order to deliver data. The comparison with a contention-free and a contention-based baselines shows that our framework achieves a superior performance in terms of goodput even in high traffic situations while maintaining a low collision rate. The scalability of the proposed method is studied, since it is a major problem in MARL and this paper provides the first results in order to address it., Comment: Paper accepted at VTC 2022 Spring Workshops. arXiv admin note: substantial text overlap with arXiv:2108.07144

Optical communications networks include basic components such as EDFAs and WSSs that present a polarization gain imbalance. The gain imbalance at the component output depends on the incident state-of-polarization. This well-known phenomenon called Polarization Dependent Loss (PDL) is revisited in this article and systematically analyzed within the framework of communications theory and practical achievable information rates. The rate impairment induced by PDL is discussed on an optical channel with distributed PDL and on a simplified equivalent lumped PDL channel. We have previously designed the New Spatially Balanced (NSB) signaling, an optimal 4D modulation based on this simplified model. We explore in this article its performance through numerical simulations and experimentally over a transmission in a recirculating loop with in-line PDL. Experimental measurements confirm an accurate channel modeling and significant information rate gains brought by NSB that results in SNR gains of the order of 0.4 dB at a practical transmission outage.

International audience; We report on the use of semiconductor optical amplifiers (SOAs) to extend the optical bandwidth of next generation optical systems to 100 nm and beyond. After discussing the technological progress and the motivation for rekindling the interest in SOAs for line amplification, we describe the innovative approach developed for the realization of ultra-wideband (UWB) SOAs. Leveraging custom design of singly polarized SOAs to provide gain over 100+ nm bandwidth, we developed a polarization diversity architecture to realize UWB-SOA modules. Embedded in a compact package, the UWB amplifier modules have been successfully used to demonstrate 100+ Tb/s transmissions. We subsequently review recent experimental transmission results based on such novel 100+ nm wide semiconductor optical amplifiers, including our first demonstration of 100+Tb/s transmission over 100 km distance, our field trial using real-time traffic, and finally the transmission of 107 Tb/s throughput over three spans of standard single mode fiber (SSMF) using hybrid UWB Raman/SOA amplification technique.

International audience; We consider three-tier network architecture modeled with two physical nodes in tandem where an autonomous agent controls the number of active resources on each node. We analyse the learning of auto-scaling strategies in order to optimise both performance and energy consumption of the whole system. We compare several model-based reinforcement learning with model-free Q-learning algorithm. The relevance of these algorithms is to faster update Q-value function with an additional planning phase allowed by approximated model of the dynamics of the environment. Secondly, we consider the same tandem queue scenario with MMPP (Markov modulated Poisson process) for arrivals. In this context, the arrival rate is varying over time and this information is hidden to the agent. Our goal is to assess the robustness of such model-based reinforcement learning algorithms in this particular scenario.

International audience; Detecting faults in distributed network systems is challenging because of their complexity, but this is required to evaluate and improve their reliability. This paper proposes ChaT, a testing and evaluation methodology under system reconfigurations and perturbations for distributed network systems, to evaluate QoS reliability by discriminating safe and failure-prone behaviors from different testing scenarios. Motivated by metamorphic testing technique that removes the burden of defining software oracles, we propose some metamorphic relationships that correlate system inputs and outputs to find patterns in executions. Classification techniques based on machine learning (principal component analysis and support vector machine) are used to identify system states and validate the proposed metamorphic relationships. These metamorphic relationships are also used to help anomaly detection. We verify this with several anomaly detection techniques (isolation forest, one-class SVM, local outlier factor, and robust covariance) that categorize experiments belonging to either safe or failure-prone states. We apply ChaT to a video streaming application use case. The simulation results show the effectiveness of ChaT to achieve our goals: identifying execution classes and detecting failure-prone experiments based on metamorphic relationships with high level of statistical scores.

In modern cloud data centers, reconfigurable devices (FPGAs) are used as an alternative to Graphics Processing Units to accelerate data-intensive computations (e.g., machine learning, image and signal processing). Currently, FPGAs are configured to execute fixed workloads, repeatedly over long periods of time. This conflicts with the needs, proper to cloud computing, to flexibly allocate different workloads and to offer the use of physical devices to multiple users. This raises the need for novel, efficient FPGA scheduling algorithms that can decide execution orders close to the optimum in a short time. In this context, we propose a novel scheduling heuristic where groups of tasks that execute together are interposed by hardware reconfigurations. Our contribution is based on gathering tasks around a high-latency task that hides the latency of tasks, within the same group, that run in parallel and have shorter latencies. We evaluated our solution on a benchmark of 37500 random workloads, synthesized from realistic designs (i.e., topology, resource occupancy). For this testbench, on average, our heuristic produces optimum makespan solutions in 47.4% of the cases. It produces acceptable solutions for moderately constrained systems (i.e., the deadline falls within 10% of the optimum makespan) in 90.1% of the cases.

Demo paper; International audience; Network slicing is one of the key building blocks in the evolution towards "zero touch networks". Indeed, this will allow 5G and beyond 5G networks to deploy services dynamically, on the same substrate network, regardless of their constraints. In this demo, we introduced a platform for dynamic virtual network embedding, a problem class known to be NP-hard. The proposed solution is based on a combination of a deep reinforcement learning strategy and a Monte Carlo (MC) approach. The idea here is to learn to generate, using a Deep Q-Network (DQN), a distribution of the placement solution, on which a MC-based search technique is applied. This makes the agent's exploration of the solution space more efficient.

International audience; Machine learning (ML) can be used in various ways to improve multi-user multiple-input multiple-output (MU-MIMO) receive processing. Typical approaches either augment a single processing step, such as symbol detection, or replace multiple steps jointly by a single neural network (NN). These techniques demonstrate promising results but often assume perfect channel state information (CSI) or fail to satisfy the interpretability and scalability constraints imposed by practical systems. In this paper, we propose a new strategy which preserves the benefits of a conventional receiver, but enhances specific parts with ML components. The key idea is to exploit the orthogonal frequency-division multiplexing (OFDM) signal structure to improve both the demapping and the computation of the channel estimation error statistics. Evaluation results show that the proposed ML-enhanced receiver beats practical baselines on all considered scenarios, with significant gains at high speeds.

International audience; We propose a per-packet distributed monitoring plane for time-sensitive optical networking. Over a packet switched network and using our FPGA-based prototype, we demonstrate latency measurements of 6.4ns precision per hop with an offset of only 1.22µs.

International audience; Network Function Virtualization (NFV) and service orchestration simplify the deployment and management of network and telecommunication services. The deployment of these services requires, typically, the allocation of Virtual Network Function-Forwarding Graph (VNF-FG), which implies not only the fulfillment of the service's requirements in terms of Quality of Service (QoS), but also considering the constraints of the underlying infrastructure. This topic has been well-studied in existing literature, however, its complexity and uncertainty of available information unveil challenges for researchers and engineers. In this paper, we explore the potential of reinforcement learning techniques for the placement of VNF-FGs. However, it turns out that even the most well-known learning technique is ineffective in the context of a large-scale action space. In this respect, we propose approaches to find out feasible solutions while improving significantly the exploration of the action space. The simulation results clearly show the effectiveness of the proposed learning approach for this category of problems. Moreover, thanks to the deep learning process, the performance of the proposed approach is improved over time.

Newly introduced polarization diversity probing codes are suggested to enhance the sensitivity and bandwidth performance of differential phase-sensitive distributed OTDR systems. This was recently demonstrated by means of short-length specialized fibers with a backscattering induced by equally spaced Fiber Bragg Gratings inserted in the fiber itself. The work summarized in this letter aims to extend the polarization-diversity probing technique to widely spread standard single mode fibers (SSMF) used for telecommunications, by solely exploiting the Rayleigh backscattering. Conditions to achieve perfect phase estimation along such fibers are first detailed. An experimental validation highlights the ability to detect and localize, over 25km of SSMF, multiple low-energy perturbations within a 475Hz-bandwidth.

International audience; Introduction. The Internet of Things consists of devices and software interacting altogether in order to build powerful and added-value services. One of the main challenges in this context is to support end users with simple, user-friendly, and automated techniques to design such applications. IFTTT-style rules are a popular way to build IoT applications as it addresses this challenge. Problem statement. Given the dynamicity of IoT applications, these techniques should also consider that these applications are in most cases not built once and for all. They can evolve over time and objects may be added or removed for several reasons (replacement, loss of connectivity, upgrade, failure, etc.). There is a need for techniques and tools supporting the reconfiguration of rule-based IoT applications to ensure certain correctness properties during this update tasks. Methodology. In this paper, we propose new techniques for supporting the reconfiguration of running IoT applications, represented as a set of coordinated rules acting on devices. These techniques compare two versions of an application (before and after reconfiguration) to check if several functional and quantitative properties are satisfied. This information can be used by the user to decide whether the actual deployment of the new application should be triggered or not. Contributions and results. The analysis techniques have been implemented using encodings into formal specification languages and verification is carried out using corresponding analysis frameworks. All these techniques for designing new applications, analyzing the aforementioned reconfiguration properties, and deploying the new applications have been integrated into the WebThings platform and applied on real-world examples for validation of the approach.

International audience; Cet article présente une méthode de captation de vibrations à partir de fibre optiques, inspirée des télécommunications optiquesà très haut débit et longue distance. Nous y détaillons le principe du cohérent-MIMO : un capteurà diversité de polarisation avec source laser cohérente, qui permet la captation de phénomènes mécaniques large bande.

International audience; Nous proposons une méthode pour estimer l'amplitude d'une chute de puissance dans un réseau fibré en utilisant un estimateur de profils longitudinaux de puissance basé sur l'analyse de signal reçu. Celle-ci s'appuie notamment sur une calibration et un formalisme simple. Nous analysons expérimentalement la précision de l'estimation en fonction de la distance à l'amplificateur précédent.

International audience; Network slicing has emerged as a pivotal concept in 5G systems, allowing mobile operators to build isolated logical networks (slices) on top of shared infrastructure networks. Within a network slice, several Service Function Chains are usually deployed on a best-effort premise. Nevertheless, this approach does not guarantee the availability of enough infrastructure resources to accommodate the uncertain and timevarying slice resource demands. This paper investigates two adaptive slice resource provisioning methods accounting for the evolution with time of the slice resource demands. A probabilistic guarantee of meeting the slice resource requirements can be obtained, while being robust against uncertainties. The myopic approach accounts for the past demands when provisioning the current demands, while the foresighted approach accounts for both past and future demands. These two methods lead to MILP problems. Their performance is compared with a quasi-static method, where provisioning is agnostic of the past and future demands. Index Terms-Network slicing, foresighted provisioning.

International audience; In this paper, we make an experimental comparison of semi-parametric (Cox proportional hazards model, Aalen additive model), parametric (Weibull AFT model), and machine learning methods (Random Survival Forest, Gradient Boosting Cox proportional hazards loss, DeepSurv) through the IPEC score on three different datasets (PBC, GBCSG2 and TLCM).

International audience; Network Slicing (NS) is a key technology that enables network operators to accommodate different types of services with varying needs on a single physical infrastructure. Despite the advantages it brings, NS raises some technical challenges, mainly ensuring the Service Level Agreements (SLA) for each slice. Hence, monitoring the state of these slices will be a priority for ISPs. However, due to the high measurements overhead, it is generally forbidden to directly measure the performance of all of these slices. To overcome this limitation, network tomography is a promising solution, consisting of a set of methods of inferring unmeasured network metrics using end-to-end measurements between monitors. In this work, we focus on inferring the additive metrics of slices such as delays or logarithms of loss rates. We model the inference task as a regression problem that we solve using neural networks. In our approach, we train the model on an artificial dataset. This not only avoids the costly process of collecting a large set of labeled data but has also a nice covering property useful for the procedure's accuracy. Moreover, to handle a change on the topology or the slices we monitor, we propose a solution based on transfer learning in order to find a trade-off between the quality of the solution and the cost to get it. Simulation results with both, emulated and simulated traffic show the efficiency of our method compared to existing ones in terms of both accuracy and computation time.

International audience; We demonstrate the performance of a novel multi-carrier MIMO fiber sensing interrogator based on OFDM approach. Over an installed telecom fiber cable, coherent fading is mitigated by means of a purely digital subcarrier combiner.

International audience; Network slicing remains one of the key technologies in 5G and beyond 5G networks (B5G). By leveraging SDN and NVF techniques, it enables the coexistence of several heterogeneous virtual networks (VNs) on top of the same physical infrastructure. Despite the advantages it brings to network operators, network slicing raises a major challenge: Resource allocation of VNs, also known as the virtual network embedding problem (VNEP). VNEP is known to be an NP-Hard problem. Several heuristics, meta-heuristics and Deep Reinforcement Learning (DRL) based solutions were proposed in the literature to solve it. Regarding the first two categories, they can provide a solution for large scale problems within a reasonable time, but the solution is usually suboptimal, which leads to an inefficient utilization of the resources and increases the cost of the allocation process. For DRL-based approaches and due to the explorationexploitation dilemma, the solution can be infeasible. To overcome these issues, we combine, in this work, deep reinforcement learning and relational graph convolutional neural networks in order to automatically learn how to improve the quality of VNEP heuristics. Simulation results show the effectiveness of our approach. Starting with an initial solution given by the heuristics our approach can find an amelioration, with an improvement in the order of 35%.

International audience; Orthogonal frequency-division multiplexing (OFDM) is widely used in modern wireless networks thanks to its efficient handling of multipath environment. However, it suffers from a poor peak-to-average power ratio (PAPR) which requires a large power backoff, degrading the power amplifier (PA) efficiency. In this work, we propose to use a neural network (NN) at the transmitter to learn a high-dimensional modulation scheme allowing to control the PAPR and adjacent channel leakage ratio (ACLR). On the receiver side, a NN-based receiver is implemented to carry out demapping of the transmitted bits. The two NNs operate on top of OFDM, and are jointly optimized in and end-to-end manner using a training algorithm that enforces constraints on the PAPR and ACLR. Simulation results show that the learned waveforms enable higher information rates than a tone reservation baseline, while satisfying predefined PAPR and ACLR targets.

International audience; We present low complexity and efficient path selection algorithms for limiting the exposure to pandemic virus during and between lock-downs. The average outdoor exposure rate reduction is around 3.

Differential phase optical time-domain reflectometry is subject to Rayleigh fading, which locally impedes proper phase extraction. We introduce in this Letter a reliability metric derived from the measured Jones matrix estimates of the fiber sensor to discriminate between true perturbations and false alarms caused by Rayleigh backscatter fading points. The proposed soft-decision metric is also of interest for further post-processing stages exploiting artificial intelligence based classifiers.

International audience; VLC based indoor positioning system is a natural choice thanks to LED technology and future trend. We can reuse existing LED lighting installations to design an appropriate scheme to provide indoor localization. By receiving basic information such as the light intensity and knowing the position of the transmitters for example through Visible Light Communication (VLC), one can estimate its position with the potential of high accuracy. In this paper, we use a multi-color sensor and build an indoor localization system based on the light fingerprint, which is suitable for smart cities and homes, health-care centers, hospitals or similar. There are, however, several practical problems that make it a technology that is not yet sufficiently developed to be used in our everyday life. In this paper, a new perspective is given by looking into indoor localization using multi-color fingerprinting and different machine learning methods. Experiment results have shown its effectiveness and potentials with mean localization errors of around ten centimeters.

International audience; Cloud applications and services have significantly increased the importance of system and service configuration activities. These activities include updating(i) these services, (ii) their dependencies on third parties,(iii) their configurations, (iv) the configuration of the execution environment, (v) network configurations. The high frequency of updates results in significant configuration complexity that can lead to failures or performance drops.To mitigate these risks, service providers extensively rely on testing techniques, such as metamorphic testing, to detect these failures before moving to production. How-ever, the development and maintenance of these tests are costly, especially the oracle, which must determine whether a system’s performance remains within acceptable boundaries. This paper explores the use of a learning method called Principal Component Analysis (PCA) to learn about acceptable performance metrics on cloud-native services and identify a metamorphic relationship between the nominal service behavior and the value of these metrics. We investigate the following research question: Is it possible to combine the metamorphic testing technique with learning methods on service monitoring data to detect error-prone reconfigurations before moving to production? We remove the developers’ burden to define a specific oracle in detecting these configuration issues. For validation, we applied this proposal on a distributed media streaming application whose authentication was managed by an external identity and access management services.This application illustrates both the heterogeneity of the technologies used to build this type of service and its large configuration space. Our proposal demonstrated the ability to identify error-prone reconfigurations using PC

With network slicing in 5G networks, Mobile Network Operators can create various slices for Service Providers (SPs) to accommodate customized services. Usually, the various Service Function Chains (SFCs) belonging to a slice are deployed on a best-effort basis. Nothing ensures that the Infrastructure Provider (InP) will be able to allocate enough resources to cope with the increasing demands of some SP. Moreover, in many situations, slices have to be deployed over some geographical area: coverage as well as minimum per-user rate constraints have then to be taken into account. This paper takes the InP perspective and proposes a slice resource provisioning approach to cope with multiple slice demands in terms of computing, storage, coverage, and rate constraints. The resource requirements of the various SFCs within a slice are aggregated within a graph of Slice Resource Demands (SRD). Infrastructure nodes and links have then to be provisioned so as to satisfy all SRDs. This problem leads to a Mixed Integer Linear Programming formulation. A two-step approach is considered, with several variants, depending on whether the constraints of each slice to be provisioned are taken into account sequentially or jointly. Once provisioning has been performed, any slice deployment strategy may be considered on the reduced-size infrastructure graph on which resources have been provisioned. Simulation results demonstrate the effectiveness of the proposed approach compared to a more classical direct slice embedding approach., 15 pages, 13 figures. Parts of this work have been presented at IEEE GLOBECOM 2018 (Abu Dhabi, UAE, https://ieeexplore.ieee.org/abstract/document/8648039) and IEEE ICC 2020 (Dublin, Ireland, https://ieeexplore.ieee.org/document/9148897)

L’idée développée dans cette thèse est d’utiliser le Machine Learning/Deep Learning et l’analyse de données radio 3GPP pour estimer et prédire le comportement d’un utilisateur, en termes d’habitudes et de préférenced’usage des services mobiles d’un réseau 5G. Le caractère multidimensionnel du comportement de l’utilisateur rend son estimation complexe et reste actuellement un défi. On a donc étudié son estimation sous une approche innovante au regard de l’état de l’art. On a proposé de la réaliser au sein d’un système unifié qui estime en parallèle chaque dimension du comportement. En utilisant des méthodes basées sur l’apprentissageapprofondi (deep-learning) supervisé et hybride/semi-supervisé, on propose une solution pour la détection de l’environnement (Indoor/Outdoor Detection (IOD)) et jusqu’à 8 classes d’environnement d’un utilisateur de téléphone portable. Nous proposons ensuite une solution permettant de détecter la catégorie de mobilité (Mobility Speed Profile (MSP) Detection) jusqu’à 8 profils de vitesses. Enfin, une solution innovante basée sur des algorithmes d’apprentissage profond dans une architecture multitâches permet d’estimer conjointement à la fois l’environnement et le profil de mobilité. La comparaison avec l’état de l’art a montré l’efficacité desméthodes proposées. Ce qui permet d’envisager leur utilisation par des opérateurs mobiles dans le futur., The main idea of this thesis is to use machine learning/deep learning techniques to estimate and predict user behavior by analyzing 3GPP radio signals. The user behavior is defined in terms of habits and preferences while consuming 5G services. The estimation of user behavior is complex and remains a challenge due to its multidimensional nature. We therefore studied an innovative approach for the user behavior estimation: we use a unified system which jointly as well as parallelly estimates each dimension of behavior. Using methods basedon supervised and hybrid / semi-supervised deep-learning, we propose a solution for the detection of the user environment (from Indoor / Outdoor Detection (IOD) to up to 8 classes). We then propose a solution to detectthe mobility categories (Mobility Speed Profile (MSP) Detection) up to 8 speed profiles. Finally, an innovative solution jointly estimates both the environment and the mobility profile using deep learning algorithms and a multitasking architecture. The comparison with the state of the art shows the effectiveness of the proposed methods. This allows to consider its deployment by operators in the future.

International audience; Searching for documents is a task that everyone faces on a regular basis, especially when looking for a relevant Internet page, an e-mail, or a document on an Intranet. An effective search relies on a precise and well-organized search engine.The majority of current techniques combine a keyword search with structural information (ontologies, relationships between elements) in order to order the documents in a corpus by relevance.In this article, we present a new navigation engine called Gismo (Generic Information Search with a Mind of its Own). Gismo exploits only the textual content of documents and does not require ontology, metadata, or prelearning. It is thus possible to use it on any corpus without making assumptions about the type of documents or their language. The model chosen and the algorithms used allow Gismo to be extremely fast even on large corpora. Finally, Gismo allows you to find, sort and organize documents by theme and relevance, making it an navigation engine and not a simple search engine.; La recherche de documents (pages Web, courriels, fichiers Intranet\...) est une tâche à laquelle tout un chacun est confronté régulièrement. Une recherche efficace repose sur un moteur de recherche précis et bien organisé.La majorité des techniques actuelles combinent une recherche par mots-clés à des informations structurelles (ontologies, relations entre éléments) afin de classer les documents d'un corpus par pertinence.Nous présentons dans cet article un nouveau moteur de navigation, appelé Gismo (Generic Information Search with a Mind of its Own). Gismo exploite uniquement le contenu textuel des documents et ne nécessite ni ontologie, ni méta-données, ni pré-apprentissage. Il est ainsi possible de l'utiliser sur n'importe quel corpus sans faire d'hypothèse sur le type ou la langue des documents. Le modèle choisi et les algorithmes utilisés permettent à Gismo d'être extrêmement rapide même sur des grands corpus. Gismo permet de trouver, trier et organiser les documents par thème et par pertinence, ce qui en fait un moteur de navigation et non un simple moteur de recherche.

International audience; Searching for documents is a task that everyone faces on a regular basis, especially when looking for a relevant Internet page, an e-mail, or a document on an Intranet. An effective search relies on a precise and well-organized search engine.The majority of current techniques combine a keyword search with structural information (ontologies, relationships between elements) in order to order the documents in a corpus by relevance.In this article, we present a new navigation engine called Gismo (Generic Information Search with a Mind of its Own). Gismo exploits only the textual content of documents and does not require ontology, metadata, or prelearning. It is thus possible to use it on any corpus without making assumptions about the type of documents or their language. The model chosen and the algorithms used allow Gismo to be extremely fast even on large corpora. Finally, Gismo allows you to find, sort and organize documents by theme and relevance, making it an navigation engine and not a simple search engine.; La recherche de documents (pages Web, courriels, fichiers Intranet\...) est une tâche à laquelle tout un chacun est confronté régulièrement. Une recherche efficace repose sur un moteur de recherche précis et bien organisé.La majorité des techniques actuelles combinent une recherche par mots-clés à des informations structurelles (ontologies, relations entre éléments) afin de classer les documents d'un corpus par pertinence.Nous présentons dans cet article un nouveau moteur de navigation, appelé Gismo (Generic Information Search with a Mind of its Own). Gismo exploite uniquement le contenu textuel des documents et ne nécessite ni ontologie, ni méta-données, ni pré-apprentissage. Il est ainsi possible de l'utiliser sur n'importe quel corpus sans faire d'hypothèse sur le type ou la langue des documents. Le modèle choisi et les algorithmes utilisés permettent à Gismo d'être extrêmement rapide même sur des grands corpus. Gismo permet de trouver, trier et organiser les documents par thème et par pertinence, ce qui en fait un moteur de navigation et non un simple moteur de recherche.

International audience; Design Space Exploration of data-flow Systems-on-Chip either focuses on classical shared bus or on complex network-on-chip (NoC) architectures. A lack of research work exists that targets segmented bus architectures. These offer performance improvements (latency, power consumption) with respect to a shared bus, while employing much simpler communication structures and algorithms than a NoC. Despite the lack in the research work, segmented buses are popular in multi-processor systems and in FPGA interconnects. This paper fills this lack with two contributions. First, we propose a Satisfiability Modulo Theory (SMT) formulation. Secondly, we provide a technique to reduce the design-space explosion problem that is portable to other formulations (e.g., ILP, MILP) and to problems where the scheduling on units (e.g., bus, CPU) is multiplexed in time. We integrated these contributions in a state-of-the-art design tool that we employ for evaluation purposes with a set of streaming applications and a MPSoC platform. The resulting framework can study the performance of fixed interconnects as well as determine the optimal architecture among a set of candidates. Our reduction technique improves considerably the scalability of DSE. For our testbench, we reduce the SMT solver run-time from 20 up to 589 times.

International audience; One of the key features of small-world networks is the ability to route messages in a few hops, using a decentralized algorithm in which each node has a limited knowledge of the topology. In 2000, Kleinberg proposed a model based on an augmented grid that asymptotically exhibits such a property. In this paper, we revisit the original model with the help of numerical simulations. Our approach is fueled by a new algorithm that can sample augmenting links in an almost constant time. The speed gain allows us to perform detailed numerical evaluations. We first observe that, in practice, the augmented scheme proposed by Kleinberg is more robust than what is predicted by the asymptotic behavior, even in very large finite grids. We also propose tighter bounds on the asymtotic performance of Kleinberg's greedy routing algorithm. We finally show that, if the model is fed with realistic parameters, the results are in line with real-life experiments.

International audience

International audience; In telecommunications, fault management is critical to improve network availability and user experience. To enhance reliability of their networks, operators require tools to quickly understand the cause of an outage. In particular, logs of alarms keep track of failures arising in their infrastructures. Due to the increasing size of networks and to the high diversity of technologies, these files may be verbose and noisy. That is why analyzing a log is often complex, hence delaying recovery and then degrading network availability. This demo presents a tool suite dedicated to log analysis. Our methodology is illustrated through the processing of a real alarm log issued from a 4G network. First, one can simplify the log by discarding irrelevant alarms and by clustering co-occurrent ones. Then, the underlying graph structure, called DIG-DAG, can store causal patterns by processing the input log online. Hence, experts can query the DIG-DAG to retrieve small and interpretable patterns.

We propose a novel interrogation technique for Distributed Acoustic Sensing (DAS) and demonstrate the detection and recognition of multiple vibration events over 50 km of SSMF. A differential-phase optical time-domain reflectometry ( $\Delta \phi -$ OTDR) approach is used. At the transmitter side, a continuous ultra-narrow-linewidth laser is modulated with polarization-multiplexed orthogonal binary sequences and at the receiver side, the back-scattered signal is captured through a polarization-diversity coherent mixer in a self-homodyne configuration. The detection of the full optical field vector and the use of an ultra-narrow-linewidth laser source enhance the sensitivity of the interrogating solution. A comparison of the experimental results to a simulation model is developed to better understand the noise limits. A low-complexity localization technique based on a multi-resolution approach is also presented. Major results include the identification of an engine noise, paving the way for numerous smart-city/industrial monitoring applications over deployed telecom fibres.

This paper deals with polarization-multiplexed optical transmissions. It addresses the challenge of designing optimal yet practical modulation codes for the 2×2 MIMO fiber channel. This specific use case diverges from classical MIMO models encountered, e.g., in the wireless literature as, here, non-ergodicity and low-complexity are key. While rather simple from a fundamental viewpoint, the proposed modulation schemes appears to be optimal and geometrically pleasing as the second minimum distance turns out to be the object of interest and the associated vectors encode the basis of a regular tetrahedron in the affine space. From an operational viewpoint, these new schemes for single channel use form polarization codes that achieve unit dB gains while exhibiting low implementation complexity.

Network slicing allows Mobile Network Operators to split the physical infrastructure into isolated virtual networks (slices), managed by Service Providers to accommodate customized services. The Service Function Chains (SFCs) belonging to a slice are usually deployed on a best-effort premise: nothing guarantees that network infrastructure resources will be sufficient to support a varying number of users, each with uncertain requirements. Taking the perspective of a network Infrastructure Provider (InP), this paper proposes a resource provisioning approach for slices, robust to a partly unknown number of users with random usage of the slice resources. The provisioning scheme aims to maximize the total earnings of the InP, while providing a probabilistic guarantee that the amount of provisioned network resources will meet the slice requirements. Moreover, the proposed provisioning approach is performed so as to limit its impact on low-priority background services, which may co-exist with slices in the infrastructure network. A Mixed Integer Linear Programming formulation of the slice resource provisioning problem is proposed. Optimal joint and suboptimal sequential solutions are proposed. These solutions are compared to a provisioning scheme that does not account for best-effort services sharing the common infrastructure network., Submitted to IEEE Transactions on Network and Service Management

With network slicing, Mobile Network Operators can accommodate on a common network infrastructure various customized services from Service Providers (SPs). Usually, the Service Function Chains belonging to a slice are deployed on a best-effort basis. Nothing ensures that enough infrastructure resources can be allocated to satisfy the demands of SPs. This paper introduces a radio resources provisioning approach to satisfy the demands of slices with radio coverage constraints. By provisioning, we ensure that enough resources are reserved for further SFC deployment. Numerical results show the effectiveness of the proposed provisioning framework for a slice deployment on a mobile network infrastructure satisfying a minimum data rate for users in the geographical areas where services have to be made available.

Nowadays, long distance optical fibre transmission systems use polarization diversity multiplexed signals to enhance transmission performance. Distributed acoustic sensors (DAS) use the same propagation medium ie. single mode optical fibre, and aims at comparable targets such as covering the highest distance with the best signal quality. In the case of sensors, a noiseless transmission enables to monitor a large quantity of mechanical events along the fibre. This paper aims at extending the perspectives of DAS systems with regard to technology breakthroughs introduced in long haul transmission systems over the last decade. We recently developed a sensor interrogation method based on coherent phase-sensitive optical time domain reflectometry ($\phi$-OTDR), with dual polarization multiplexing at the transmitter and polarization diversity at the receiver. We name this technique Coherent-MIMO sensing. A study is performed from a dual-polarization numerical model to compare several sensor interrogation techniques, including Coherent-MIMO. We demonstrate that dual-polarization probing of a fibre sensor makes it insensitive to polarization effects, decreases the risks of false alarms and thus strongly enhances its sensitivity. The simulations results are validated with an experiment, and finally quantitative data are given on the performance increase enabled by Coherent-MIMO sensing., Comment: 14 pages, 20 figures and subfigures. Submitted to Optics Express (after 1st revision)

International audience; The Internet of Things (IoT) aims at sensing and altering our surrounding environment through connected objects to improve everyday life. IoT applications are built using interconnected objects with a goal to provide added-value services. However, there are still challenges in providing a secure, robust and easy-to-use end-user platform for development of such applications. In this paper, we present a end-user tool for supporting the design and deployment of smart home IoT applications. The tool first provides a graphical user interface to specify an IoT application using a rule-based composition language. Automated analysis techniques can then be called for verifying that the designed application is correct (e.g., free of deadlocks). Finally, the tool provides a rule execution engine to support application deployment. The tool is built by implementing a set of components on top of Mozilla WebThings platform, which is a concrete implementation of W3C's Web of Things specification.

International audience; Optimal symbol detection for multiple-input multiple-output (MIMO) systems is known to be an NP-hard problem. Conventional heuristic algorithms are either too complex to be practical or suffer from poor performance. Recently, several approaches tried to address those challenges by implementing the detector as a deep neural network. However, they either still achieve unsatisfying performance on practical spatially correlated channels, or are computationally demanding since they require retraining for each channel realization. In this work, we address both issues by training an additional neural network (NN), referred to as the hypernetwork, which takes as input the channel matrix and generates the weights of the neural NN-based detector. Results show that the proposed approach achieves near state-of-the-art performance without the need for re-training. Index Terms—MIMO Detection, Deep Learning, Hypernetworks, spatial channel correlation

International audience; This paper investigates the joint subcarrier and power allocation problem for the downlink of a multi-carrier non-orthogonal multiple access (MC-NOMA) system. A novel three-step resource allocation framework is designed to deal with the sum rate maximization problem. In Step 1, we relax the problem by assuming each of the users can use all subcarriers simultaneously. With this assumption, we prove the convexity of the resultant power control problem and solve it via convex programming tools to get a power vector for each user; In Step 2, we allocate subcarriers to users by a heuristic greedy manner with the obtained power vectors in Step 1; In Step 3, the proposed power control schemes used in Step 1 are applied once more to further improve the system performance with the obtained sub-carrier assignment of Step 2. To solve the maximization problem with fixed subcarrier assignments in both Step 1 and Step 3, a centralized power allocation method based on projected gradient descent algorithm and two distributed power control strategies based respectively on pseudo-gradient algorithm and iterative waterfilling algorithm are investigated. Numerical results show that our proposed three-step resource allocation algorithm could achieve comparable sum rate performance to the existing near-optimal solution with much lower computational complexity and outperforms power controlled OMA scheme. Besides, a tradeoff between user fairness and sum rate performance can be achieved via applying different user power constraint strategies in the proposed algorithm.