Most modern congestion control algorithms, that aim to optimize delay and throughput, exploit more metrics than the sole packet loss congestion information. These additional metrics are mostly based on the round trip time evolution and allow congestion controls to reach better performance, in particular on wireless and cellular links as demonstrated by Copa, BBR, or REMY. Basically, these metrics allow congestion control to estimate the queuing level of the path and its evolution, to assess the presence of congestion. Actually, a good estimation of this level obviously prevents congestion losses, but also allows assessing a ratio of error link losses among the whole observed losses. The consistency and accuracy of these metrics are key to good congestion control performance, and this explains, for instance, the good performance of Copa currently in production at Facebook. However, these metrics remain challenging and the quest of an accurate and practical estimation seems complex. This paper investigates how a novel deep learning algorithm, known as Attention, can help in assessing queuing evolution and status on an end-to-end path. Among others, we focus on the evolution of the total time spent by packets in the buffers, which is the key metric of Copa. The results unequivocally demonstrate a better accuracy of this metric used by Copa.

As a multidimensional extension of matrices, tensors (≥3D) are a natural tool for representing and processing multidimensional data arrays. Capturing and recovering this multidimensional data is a long-term challenge in image processing and related fields. In particular, four-dimensional (4D) spectral videos contain highly redundant information across the spatial (2D), spectral (1D) and temporal (1D) axes which can be exploited through a data-learned sparse basis or dictionary. However, in compressive spectral video acquisition (where the data is compressed), tackling dictionary learning is time-consuming since it increases the computational complexity and presents drawbacks for real-time processing, where offline learning is required. In this talk, I will briefly introduce tensor representation and decomposition, and its application on spectral videos in a compressive sensing scenario. Then, I will present an approach to exploit tensor sparse representation for jointly learning the transform basis and the recovering from compressed measurements of a spectral video. I will show some results of the performance of the developed framework compared with matrix-based recovery approaches, including dictionary learning.

Dark Matter is a very active field of research in modern particle physics both on the theory and the experimental side. In this seminar, I will give a pedagogical introduction to particle Dark Matter physics and sketch the main challenges of this area of research. I will first present the astrophysical and cosmological evidence for Dark Matter and its general properties that can be inferred from observation. I will then move to the field of particle physics and discuss the general requirements of a viable Dark Matter model as well as the on-going experimental efforts to detect a potential Darl Matter particle. I will finish by showing a specific Dark Matter model, the supersymmetric WIMP, on which I have worked during my thesis.

Missing data is a recurrent problem in remote sensing, mainly due to cloud coverage for multispectral images and acquisition problems. This can be a critical issue for crop monitoring, especially for applications relying on machine learning techniques, which generally assume that the feature matrix does not have missing values. This paper proposes a Gaussian Mixture Model (GMM) for the reconstruction of parcel-level features extracted from multispectral images. A robust version of the GMM is also investigated, since datasets can be contaminated by inaccurate samples or features (e.g., wrong crop type reported, inaccurate boundaries, undetected clouds, etc). Additional features extracted from Synthetic Aperture Radar (SAR) images using Sentinel-1 data are also used to provide complementary information and improve the imputations. The robust GMM investigated in this work assigns reduced weights to the outliers during the estimation of the GMM parameters, which improves the final reconstruction. These weights are computed at each step of an Expectation-Maximization (EM) algorithm by using outlier scores provided by the isolation forest (IF) algorithm. Experimental validation is conducted on rapeseed and wheat parcels located in the Beauce region (France). Overall, we show that the GMM imputation method outperforms other reconstruction strategies. A mean absolute error (MAE) of 0.013 (resp. 0.019) is obtained for the imputation of the median Normalized Difference Index (NDVI) of the rapeseed (resp. wheat) parcels. Other indicators (e.g., Normalized Difference Water Index) and statistics (for instance the interquartile range, which captures heterogeneity among the parcel indicator) are reconstructed at the same time with good accuracy. In a dataset contaminated by irrelevant samples, using the robust GMM is recommended since the standard GMM imputation can lead to inaccurate imputed values. An application to the monitoring of anomalous crop development in the presence of missing data is finally considered. In this application, using the proposed method leads to the best detection results, especially when SAR data are used jointly with multispectral images. Exploiting the information contained in cloudy multispectral images instead of removing these images is beneficial for this application.

Satellite constellations have taken a new impetus in recent years with even more ambitious future rojects. The momentum has lasted long enough to raise the interest of the research community in addressing the adequacy of protocols mainly designed and used for terrestrial networks, to these satellite communications. There are thus versions of TCP specially designed for satellite networks [1]-[5]. Nevertheless, these previous works could turn out to be obsolete, in particular due to the recent versions of TCP based, for instance, on hybrid-type congestion control algorithms. The question we tackled in this thesis is : are the recent versions of TCP, such as CUBIC and BBR, able to meet the needs in such an environment ? We evaluated the differences between past and currently deployed TCP stacks. We gave an overview of the evolution of the use of protocols from a transport layer point of view of CUBIC and BBR. We identified the sources of delay variation in satellite constellations in order to study their impact and frequency. Modern variants of TCP accommodate this, especially for long flows. We then looked at the fairness between the flows with or without different variants. We highlighted some levels of unfairness in the heterogeneous contexts that are fairly consistent with those found in the terrestrial context. All of these studies were conducted through discrete event simulations but also emulations in order to obtain more realistic results.

Satellite constellations have taken a new impetus in recent years with even more ambitious future rojects. The momentum has lasted long enough to raise the interest of the research community in addressing the adequacy of protocols mainly designed and used for terrestrial networks, to these satellite communications. There are thus versions of TCP specially designed for satellite networks [1]-[5]. Nevertheless, these previous works could turn out to be obsolete, in particular due to the recent versions of TCP based, for instance, on hybrid-type congestion control algorithms. The question we tackled in this thesis is : are the recent versions of TCP, such as CUBIC and BBR, able to meet the needs in such an environment ? We evaluated the differences between past and currently deployed TCP stacks. We gave an overview of the evolution of the use of protocols from a transport layer point of view of CUBIC and BBR. We identified the sources of delay variation in satellite constellations in order to study their impact and frequency. Modern variants of TCP accommodate this, especially for long flows. We then looked at the fairness between the flows with or without different variants. We highlighted some levels of unfairness in the heterogeneous contexts that are fairly consistent with those found in the terrestrial context. All of these studies were conducted through discrete event simulations but also emulations in order to obtain more realistic results.

In this paper, we propose hybrid GaN-FDSOI Front-End-Modules (FEMs) combined with lens-based mmWave antenna modules for energy-efficient multi-beamforming systems. X-topology Lattice-based balanced (LB) and unbalanced (LU) differential switches are designed and fabricated using FDSOI platforms for building scalable multi-beamforming FEMs. Unified mmWave and Baseband correlation technologies based on energy density and EVM (Error Vector Magnitude) metrics are introduced for low complexity (leveraging sparsity of MIMO channels) cost-effective multi-beamforming systems. Combination of convolutional accelerators with analog signal processing for linking beam-selection algorithms to stochastic-wave-shaping (SWS) will create new paradigms for eliminating the concept of “elements” in arrays and replacing it by the vision of “radiating current flows” over a textured surface (metasurface or metavolume). The radiating metasurfaces or metavolumes which can be conformal to the patterned optical lenses are fed by a limited number of emitting/receiving points. The use of time-modulated correlation functions will foster new system architectures with critical functionalities including direction of arrival (DOA) estimation and secure communications.

For more than three decades, Global Navigation Satellite System (GNSS) signals have been seen as signals of opportunity as in GNSS Reflectometry (GNSS-R). The study of the reflections from the ground of such signals can indeed lead to many features regarding the reflecting surface and the receiver’s height. Due to the nature of the GNSS signal, that is, due to its wavelength, the distortion of the reflected signal may vary significantly depending on the reflecting surface and on the dynamic and height of the receiver. The latter does range from low earth orbit down to ground-based platforms. In this last case, the vicinity to the ground induces important interference between the direct and the reflected path which makes it difficult to process directly in order to obtain altimetry product. In this study, the feasibility of ground-based single antenna GNSS-R altimetry is studied and solutions are presented depending on the satellite elevation angle. To do so, maximum-likelihood-based algorithms - namely the CLEAN-RELAX Estimator and the Approximate Maximum Likelihood Estimator - are presented and applied to a set of scenarios.

This paper describes new impact analyses carried out on the imperfections of clock signals in the innovative but not well-known N-path Frequency Down-converter. Using an evolution of a previously developed analytical model, these analyses give more insight on the performance of the N-path Frequency Down-converter and allow more precise trouble-shooting of an Integrated Circuit realized on this principle. They are compared with measurements.

At their design time, GPS L5 and GALILEO E5a/E5b signals compatibility with TACAN/DME was studied for aeronautical users with altitude limited up to 40,000 feet, but not for space-borne users. Aircraft would see a few strong pulses and a mitigation technique as simple as pulse blanking would usually work as mitigation means. For space-borne GNSS receivers in Low Earth Orbit (LEO), if the larger free space losses lead to weaker received TACAN/DME signals, the number of beacons in visibility is much higher, reaching in the worst locations a total over two hundred with more than half of them having a peak power above or close to the noise floor, making time blanking a poor mitigation means. Therefore, other mitigation techniques performances need to be assessed in order to determine which techniques are best suited. A simulation tool was developed to compute the post correlation C/No degradation due to TACAN/DME on a LEO with and without mitigation means enabled. The equivalent post-correlation noise (No) increase due to TACAN/DME, or what remains after a mitigation technique is applied, is simulated using the Spectral Separation Coefficient (SSC) methodology to emulate the effect of GNSS signal de-spreading in the receiver correlation process. The part of the useful signal carrier suppressed by the application of a mitigation technique (time blanking and/or frequency notch filtering) is taken into account in the simulation. This study is focusing on radio-occultation (RO) missions which are the more sensitive to TACAN/DME interferences. Indeed, a medium-gain antenna (9–18 dB typical) is steered toward the earth limb resulting in having many TACAN/DME transmitters inside its main lobe. This configuration can lead to a high received power from them, as the LEO RO satellite is also in their main antenna lobe. In this configuration, the C/No degradation, plotted on a geographic map can reach up to 13.8 dB in the absence of mitigation over the European TACAN/DME hotspot. Several promising mitigation techniques have been included in the simulation tool to determine which one shall be implemented on board a LEO RO satellite mission: time domain pulse blanking, Frequency Domain Adaptive Filtering (FDAF) or hybrid blanking. We also considered implementing pulse cancellation, an attractive technique in theory, but not so in practice due to the deviation of the actual transmitted signals with respect to their theoretical models. As anticipated, pulse blanking does not perform well at the LEO orbit. It can be actually worse than doing nothing when there is a large number of TACAN/DME transmitters in visibility since it leads to a high loss in useful GNSS signals during the blanking process. As detailed in this paper, hybrid time domain and frequency domain methods are more effective when frequency notch filtering is applied over a limited time window . For FDAF, windows have fixed boundaries, independently of the presence of interfering pulses, whereas in the hybrid method, the time windows are centered on the detected pulses. The FDAF method reduced the peak interference down to 5.6 dB. The hybrid blanking method has the best performances with a worst degradation which can be reduced to 4.3 dB over the European hotspot.