Recently, many different Random Access protocols have been developed and proposed for satellite return link communications. Synchronous and asynchronous solutions vary, mainly, in terms of signaling overhead regarding synchronization information. On the one hand, Contention Resolution Diversity Slotted Aloha (CRDSA) has emerged as a leader technique for synchronous transmissions with multiple replicas per packet and Successive Interference Cancellation at reception. On the other hand, Asynchronous Contention Resolution Diversity ALOHA (ACRDA) has been proposed as an equivalent asynchronous method to CRDSA. CRDSA and ACRDA incur a deadlock when no more packets can be retrieved due to high channel loads. Therefore, a complementary method to CRDSA: MultireplicA decoding using corRelation baSed locAlisation (MARSALA) proposed to combine replicas belonging to the same undecoded packet after localizing them through correlations. This allows to unlock some of the deadlock configurations which would relaunch CRDSA again. In asynchronous transmissions, Enhanced Contention Resolution Aloha (ECRA) uses different combining techniques for packets replicas to offer high system performance in terms of Packet Loss Ratio (PLR) and throughput. The former and latter techniques MARSALA and ECRA can be costly in localization complexity to the receiver. Therefore, Shared Position Technique for Interfered Random Transmissions (R-SPOTiT) defines a way to reduce the complexity of MARSALA’s packets localization without degrading performance nor adding extra signaling information. Accordingly, this paper proposes ARSPOTiT, an asynchronous design of R-SPOTiT, as a complementary method to ACRDA that introduces a way to locate replicas on their virtual frames with less complexity and significantly higher system performance compared to ACRDA.

Global Navigation Satellite System (GNSS) are present in our daily lives. Moreover, new users are emerging with further operation needs involving a constant evolution of the current navigation systems. In the current framework of Galileo (GNSS European system) and especially within the Galileo E1 Open Service (OS), adding a new acquisition aiding signal could contribute to provide higher resilience at the acquisition phase, as well as to reduce the time to first fix (TTFF). Designing a new GNSS signal is always a trade-off between several performance figures of merit. The most relevant are the position accuracy, the sensitivity and the TTFF. However, if one considers that the signal acquisition phase is the goal to design, the sensitivity and the TTFF have a higher relevance. Considering that, in this thesis it is presented the joint design of a GNSS signal and the message structure to propose a new Galileo 2nd generation signal, which provides a higher sensitivity in the receiver and reduce the TTFF. Several aspects have been addressed in order to design a new signal component. Firstly, the spreading modulation definition must consider the radio frequency compatibility in order to cause acceptable level of interference inside the band. Moreover, the spreading modulation should provide good correlation properties and good resistance against the multipath in order to enhance the receiver sensitivity. Secondly, the choice of the new PRN code is also crucial in order to ease the acquisition phase. A simple model criterion based on a weighted cost function is used to evaluate the PRN codes performance. This weighted cost function takes into account different figures of merit such as the autocorrelation, the cross-correlation and the power spectral density. Thirdly, the design of the channel coding scheme is always connected with the structure of the message. A joint design between the message structure and the channel coding scheme can provide both, reducing the TTFF and an enhancement of the resilience of the decoded data. In this this, a new method to co-design the message structure and the channel coding scheme for the new G2G signal is proposed. This method provides the guideline to design a message structure whose the channel coding scheme is characterized by the full diversity, the Maximum Distance Separable (MDS) and the rate compatible properties. The channel coding is essential in order to enhance the data demodulation performance, especially in harsh environments. However, this process can be very sensitive to the correct computation of the decoder input. Significant improvements were obtained by considering soft inputs channel decoders, through the Log Likelihood Ratio LLRs computation. However, the complete knowledge of the channel state information (CSI) was usually considered, which it is infrequently in real scenarios. In this thesis, we provide new methods to compute LLR approximations, under the jamming and the fading channels, considering some statistical CSI. Finally, to transmit a new signal in the same carrier frequency and using the same High Power Amplifier (HPA) generates constraints in the multiplexing design, since a constant or quasi constant envelope is needed in order to decrease the non-linear distortions. Moreover, the multiplexing design should provide high power efficiency to not waste the transmitted satellite power. Considering the precedent, in this thesis, we evaluate different multiplexing methods, which search to integrate a new binary signal in the Galileo E1 band while enhancing the transmitted power efficiency.

In May 2016, Google decided to release an Android API enabling developers and researchers to access GNSS raw data measurements from embedded GNSS chipsets. This initiative potentially allows billions of smartphone users to achieve individual precise positioning [1]. Recently, smartphones’ GNSS capabilities were optimized with the release of multi-constellation and multi-frequency GNSS chipsets such as the Broadcom 47755 in Xiaomi Mi8 [2]. However, in constrained environments, signals degradation prevents mobile users to obtain sub-metric precision especially due to the linearly polarized smartphones’ GNSS antenna. To overcome this, a smartphone collaborative positioning method will be developed and implemented, within a network of users, in order to exchange qualitative GNSS information. These exchanged data are derived from signals degradation status, quality, availability and integrity obtained via other users and abundant data from today’s urban environment. In the interest of developing this method, measurements error models must be established for different GNSS chipsets/smartphones, in an urban environment. The aim of this paper is then to draw statistical measurements models in order to characterize Android GNSS raw data measurements. Therefore, we analyzed GNSS raw data measurements obtained during a data collection campaign in Toulouse, France. Two vehicles, equipped with geodetic grade receivers, antenna and IMUs for reference purposes, were used in this campaign to collect data from seven recent smartphones (Xiaomi Mi9, Google Pixel 3, Huawei Mate 20X, 2 Honor View 20 and two Xiaomi Mi8). The structure of this article will be as follows: first, the context of smartphone collaborative positioning is presented. Thereafter, the protocol of the data collection campaign will be described in detail and different collaborative scenarios will be introduced. Finally, the statistical analysis will be made for the different scenarios and a generalized method will be extracted in order to characterize any smartphone embedded GNSS chipset’s measurements.

Le nouveau paradigme Information Centric Network (ICN), et en particulier l'architecture Named Data Networking (NDN), redéfinissent une couche réseau accès sur le contenu. En effet, les tendances d'utilisation d'Internet ont bien évoluées depuis l'invention d'IP (Youtube, Netflix, ...). Ce nouveau type de réseau promet une utilisation plus intelligente du réseau pour garantir une meilleure Qualité d’Expérience à ses utilisateurs. Mais qu'en est-il vraiment ? Après une description des principales caractéristiques de NDN, nous vous proposons de vous présenter une solution pour répartir équitablement les flux des différents utilisateurs dans ce type de réseau.

L'objectif général de la présentation est de déterminer les performances des futurs signaux 5G en terme de positionnement. Pour ce faire, l'étude se focalise sur la définition d'un modèle mathématique de sorties de corrélateurs 5G. D'après les standards 3GPP, la 5G utilisera des signaux OFDM (Orthogonal Frequency Division Multiplexing). Dans la littérature, des modèles mathématiques aux différentes étapes du traitement (modulation, démodulation, corrélation, boucle de poursuite,...) pour signaux OFDM sont déjà disponible. Ces modèles sont développés en supposant un canal de propagation constant pendant la durée d'un symbole OFDM. En vue d'adapter ces résultats, ces modèles, à la 5G, il a été nécessaire de sélectionner un canal de propagation répondant aux critères des standards 3GPP pour la 5G. Le choix s'est porté sur le simulateur QuaDRiGa ; une étude approfondie des modèles générés par ce simulateur a montré que le canal de propagation ne pouvait être considéré comme constant sur la durée d'un symbole OFDM. Ainsi, les modèles développés dans la littérature ne sont pas utilisables tels quels. Dans cette présentation, de nouveaux modèles mathématiques, prenant en compte cette évolution de canal, sont développés et appliqués pour le calcul de pseudo- distance 5G.

Single-photon lidar has emerged as a prime candidate technology for depth imaging through challenging environments. Until now, a major limitation has been the significant amount of time required for the analysis of the recorded data. Here we show a new computational framework for real-time three-dimensional (3D) scene reconstruction from single-photon data. By combining statistical models with highly scalable computational tools from the computer graphics community, we demonstrate 3D reconstruction of complex outdoor scenes with processing times of the order of 20ms, where the lidar data was acquired in broad daylight from distances up to 320metres. The proposed method can handle an unknown number of surfaces in each pixel, allowing for target detection and imaging through cluttered scenes. This enables robust, real-time target reconstruction of complex moving scenes, paving the way for single-photon lidar at video rates for practical 3D imaging applications.

Joint deconvolution and segmentation of ultrasound images is a challenging problem in medical imaging. By adopting a hierarchical Bayesian model, we propose an accelerated Markov chain Monte Carlo scheme where the tissue reflectivity function is sampled thanks to a recently introduced proximal unadjusted Langevin algorithm. This new approach is combined with a forward-backward step and a preconditioning strategy to accelerate the convergence, and with a method based on the majorizationminimization principle to solve the inner nonconvex minimization problems. As demonstrated in numerical experiments conducted on both simulated and in vivo ultrasound images, the proposed method provides high-quality restoration and segmentation results and is up to six times faster than an existing Hamiltonian Monte Carlo method.

The purpose of this paper is to introduce a customized multi- modality phantom designed to facilitate the proof-of-concept of MRI/ultrasound fusion approaches. Phantom experiments are often required before in vivo validation, giving access to more challenging data than numerical simulations. Nevertheless, manufactured phantoms are expensive and usually lack of flexibility. In contrast, the proposed model was inexpensive and accurately designed to overcome multimodal registration issues.

Estimating the cardiac motion from ultrasound (US) images is an ill-posed problem that requires regularization. In a recent study, it was shown that constraining the cardiac motion fields to be patch-wise sparse in a learnt overcomplete motion dictionary is more accurate than local parametric models (affine) or global functions (B-splines, total variation). In this work, we extend this method by incorporating temporal smoothness in a multi-frame optical-flow (OF) strategy. An efficient optimization strategy using the constrained split augmented Lagrangian shrinkage algorithm (C-SALSA) is proposed. The performance is evaluated on a realistic simulated cardiac dataset with available ground-truth. A comparison with the pairwise approach shows the interest of the proposed temporal regularization and multi-frame strategy in terms of accuracy and computational time.