La détection en quasi-temps réel (NRT pour Near Real-Time) de la perte de forêts tropicales est essentielle pour la conservation de la biodiversité et la gestion du carbone. Les systèmes actuels de surveillance par télédétection satellitaire présentent cependant des limites, liées à la sensibilité des données à la saisonnalité et la variabilité environnementale. De plus, les approches optiques souffrent souvent de longs délais de détection en raison d’une couverture nuageuse fréquente, tandis que les méthodes utilisant des données radar à synthèse d’ouverture (SAR pour Synthetic Aperture Radar) sont affectées par la variabilité du speckle et la perte de résolution causée par le filtrage des images, ce qui réduit leur sensibilité aux perturbations à petite échelle. Cette thèse présente une approche non supervisée de détection bayésienne en ligne de points de changement (BOCD pour Bayesian Online Changepoint Detection) pour la détection NRT de la perte de forêt, évaluée à l’aide de données de référence fiables MapBiomas Alerta, couvrant l’Amazonie brésilienne et le Cerrado. La méthode estime la déforestation de manière récursive en ligne selon le principe du Maximum A Posteriori (MAP). Elle se base sur la distribution a posteriori du nombre d’acquisitions effectuées depuis le plus récent changement, modélisé comme une variable aléatoire associée à l’état caché d’un modèle de Markov. Le traitement récursif et des statistiques a priori conduisent à une gestion efficace de la variabilité du signal et de la saisonnalité. La conjugaison a priori permet la mise à jour de paramètres à coût calculatoire réduit, et un mécanisme d’analyse de survie tient compte du contexte spatiotemporel des perturbations. Le BOCD est tout d’abord mis en oeuvre avec des séries temporelles SAR Sentinel-1, non filtrées pour préserver les détails spatiaux. Les résultats montrent des taux de détection plus élevés pour les clairières de petite taille que ceux obtenus avec les principaux systèmes opérationnels tels que GLAD-L, RADD et GFW, tout en maintenant de faibles taux de fausses alertes. Pour les grandes clairières, le BOCD surpasse GLAD-L dans le Cerrado et égale RADD en Amazonie, bien que GFW reste supérieur grâce à sa combinaison de produits SAR et optiques. Le BOCD est ensuite étendu aux acquisitions polarimétriques Sentinel-1 (pol-BOCD), tout en conservant une faible complexité. La combinaison des canaux VV et VH améliore la sensibilité de 10 % sur les parcelles hétérogènes et renforce la robustesse face aux différentes pratiques de déforestation, avec des taux de fausses alertes systématiquement faibles. Un troisième développement méthodologique est introduit, qui généralise la détection Bayésienne en ligne à plusieurs séries temporelles asynchrones via une approche de fusion basée sur une combinaison statistique pondérée, mise en oeuvre avec les données Sentinel-1 et Sentinel-2 (ms-BOCD). Finalement, le BOCD est appliqué à un cas d’usage particulier : la détection NRT de la perte de forêt par le feu, utilisant des données de terrain collectées lors d’un incendie survenu en septembre 2024 à Paragominas, au Brésil. Les résultats montrent un accord de 88 %, et soulignent la complémentarité entre les données SAR Sentinel-1 et multispectrales Sentinel-2 : les données optiques identifient sans ambiguïté les zones brûlées et le SAR permet des observations continues pendant les périodes nuageuses affectant les données optiques. Le cadre BOCD proposé améliore la surveillance en NRT des pertes forestières tropicales en augmentant la sensibilité au déboisement de petites parcelles, en réduisant les délais de détection par rapport aux systèmes opérationnels existants et en permettant l’intégration de données multi-sources asynchrones, sans compromettre l’efficacité calculatoire. Ces caractéristiques sont pertinentes pour le développement futur d’un système opérationnel d’alerte précoce, favorisant une surveillance de la déforestation plus rapide et fiable.

In this paper, we propose to study the potential advantages that multiple-input single-output (MISO) architectures could offer compared to the classical single-input single-output (SISO) approach for coherent free space optical (FSO) ground-to-GEO feeder links, highlighting their resilience to atmospheric turbulence and the benefits of spatial diversity. Evaluating system trade-offs under both power-limited and power-augmented scenarios, we demonstrate that MISO designs can achieve higher data rates and enable more robust optical feeder links. Choosing the best MISO configuration depends on system-level priorities, including available transmit power, optical and adaptive optics (AO) complexity, throughput-latency balance, and integration requirements. Our findings demonstrate that simple MISO architectures can effectively balance aperture size, AO requirements, interleaver depth, and transmit power, positioning them as a promising solution for efficient and resilient GEO optical feeder links.

This paper presents some theoretical results and measurements of passive intermodulation (PIM) products made on a waveguide flange contact nonlinearity with 2 and 3 carriers at different power levels. The dependence of the passive intermodulation (PIM) products power on the carrier power is the main difference between active and passive intermodulation (IM) products. For easy computation, memoryless active nonlinearities are generally modelled by polynomials or by analytical mathematical functions (e.g. hyperbolic tangent) [1, 2]. These functions are continuous, have continuous derivatives of all orders and can be approximated by their Taylor series developments, at least in “small signal” conditions where the IM power is much less than the fundamental carrier power, mathematically in a domain around the origin. In these conditions, the power of each active IM products depends on the carrier power elevated to an exponent equal to the order of the IM product, e.g. exponent 3 for order 3, exponent 5 for order 5 [1, 2]. On a dB/dB graph, the slopes of IM levels versus carrier level are equal to their order 3, 5, … This is not the case for passive IM products where the level of IM products depends on the carrier level with a slope that is different from the order, generally a non-integer value between 2 and 3 and about the same for all orders on a dB/dB graph [3 - 8]. A model based on a non-analytical power function has been proposed [9]. It has been used to predict the behavior of the passive IM products such as the level of high order of products, the nearly equal slope for all IM orders, and the decrease of a 2-carrier IM product power when a third carrier is added [9, 10]. The experiment is carried out on a nonlinear graphite material introduced between two waveguide flanges in a PIM test bench in Ku band using two low PIM triplexers. Two carriers with power up to 40 watts per carrier or three carriers with power up to 40 watts per carrier can be transmitted on the test bench. The main theoretical and experimental results are presented in this paper and validate the theoretical results: a slope of 2.4 dB/dB is measured on the third order PIM product and the levels of higher order PIM products are correctly obtained in different configurations of carriers and power by using the non-analytical model based on a power law nonlinearity with an exponent of 2.4 for order 3. Measurement with 3 carriers (with the same or different power) are compared with the theory. A particular combination of 3 carrier powers shows that some third order PIM can be nearly eliminated in concordance with the theory.

CNES, the French Space Agency, has been studying space high power radio frequency (RF) effects -Multipactor, Corona and Passive Intermodulationfor many decades, starting from J. Sombrin Multipactor theory and models [1] to ongoing activities covering TRL 1 from 1 to 7 with our collaborators from academia, agencies, and industries. This paper intends to discuss recent advances related to Multipactor analysis, and present our way towards our main objective in this field: to improve modelling and experimental Multipactor predictions and the synergy between the two. We are studying electron emission physics to enhance our models and measurement methods on dielectric materials, their TEEY 2 , charge dynamics, treatment of secondary and backscattered electrons and the impact on Multipactor predictions. We are developing SPIS 3 to create a robust Multipactor modelling tools, dealing with dielectric materials and electron sources, while considering couplings with current reference software such as CST Studio, Spark3D and ANSYS Multipaction. We are also studying multipactor mitigation techniques based on surface treatments for both conductor and dielectric materials, and RF components design innovations to deal with current trends such as miniaturisation and high performances leading to high power density hence Multipactor risks. These studies align with the European roadmap on Multipactor theme [2] and complement ESA funded activities. We share the main goal as to give our community experimental and numerical tools to get better Multipactor predictions to, in fine, improve the reliability and performances of our high power RF systems. As a national agency we also engage on community awareness, explaining to various space communities that high power effects should be a common concern, and robust analyses should be better integrated in development plans and not wait for an anomaly and/or a major satellite loss to happen.

A novel Digital-Twin technology platform is introduced for enabling system-level IC-Package-PCB-Antenna co-design, co-simulation and co-verification. The platform, based on noise and correlation-aware physics-informed behavioral modeling, integrating VISION (IVCAD) software developed by Dassault Systèmes, hosts an innovative SPICE compatible broadband RLC representation of antenna elements. The ability of the platform to account for dynamic impedance loading of antennas by multi-harmonic (MH) nonlinear RF electronics is demonstrated using energy-efficient hybrid 3D heterogeneous front-end-module technologies integrating adaptive biasing and antenna tuners (load-aware matching). The Digital-Twin technology will enable new generations of tooling (unified EDA & OTA) where classical Electromagnetic (EM) metrics (radiation pattern, noise, auto and cross-correlation functions, power-spectral density) are extended with wireless-circuit metrics (e.g., EVM, SNR, ACPR, NMSE). Unification of EDA and OTA, based on holistic Multiphysics (EM, Thermal, Mechanical) approaches, will foster new standards for joint communication and sensing at any time and from anywhere (remote ubiquitous connectivity).

This paper presents some measurements made on samples of raw materials used for antenna reflectors on communication satellites. Two theoretical results have been experimented: the first one is the power dependence of the passive intermodulation products versus the power of the carriers; the second one is the direction along which intermodulation products are radiated when the incident carriers come from different directions.

This paper addresses the propagation of acoustics or ultrasonics waves through atmosphere and the causality property. The physicists community seems to agree with the following sentence: "... empirical observation indicates that such systems are indeed causal even though the transfer function may not be a causal transform." We explain that the complex gain is not causal when not properly chosen, and that this issue can be addressed.