Thales (Portugal)

Global Navigation Satellite Systems are vulnerable to jamming and spoofing threats, compromising several critical applications. Existing detection methods based on hardware solutions (antenna array, spectrogram) are low-latency and accurate but require expensive hardware, while machine learning solutions are the most effective but require extensive training and lack adaptability. This work proposes an edge-based, statistical threat detector using crowdsourced GNSS data and fuzzy logic to integrate multiple anomaly indicators. A key feature is a C-/N0-based crowdsourcing metric. Experiments show detection precision up to 88% for jamming and 97% for spoofing, with false positive rates around 1–2% and an average detection time of 10 s.

After a brief introduction of the Knowledge Graph (KG) technology, a subfield of symbolic AI used to represent and manage semantic information, the article is devoted to quality assessment, emphasizing the importance of developing trustworthy AI in such knowledge-based systems, particularly in safety-critical applications. In this context, we remind several metrics or methods that can be applied to KGs, along with examples of their implementation in the context of digital NOTAMs illustrated by the HLIF2024 Hackathon.

The French Optical Ground Station (FrOGS) [1] is developed, by a consortium of industries: Safran Data System, OGS Technologies, Bertin ALPAO, and Airbus Defence and Space, with the support of CNES.The article presents the final bi-static architecture of the station, composed of a focal plane assembly dedicated to reception (Rx FPA), at the Cassegrain focus of the received Optical Tube Assembly (Rx OTA) and the Tx/Rx FPA at the coudé focus of the Tx OTA. Each focal plan assembly is equipped with an Adaptive Optics system. FrOGS architecture is designed to be modular and scalable, in order to cope with various orbits and laser applications. The station is currently used for two Airbus/CNES demonstrations: bi-directional communications to a GEO optical terminal (the TELEO demonstration [2]) and LEO direct to earth data repatriation of Earth Observation satellites (the LASIN demonstration [3]).FrOGS started free space optical links with TELEO in 2024, a mono-static architecture was used, emission and reception being done using a single adaptive optics bench. The new Rx OA FPA at the Cassegrain focus was qualified in July 2025 on a horizontal link and first results were obtained with TELEO in the bi-static configuration.After the launch of CO3D constellation in July 2025, LASIN demonstrator enters in its qualification phase with FrOGS.FrOGS, as an OGS demonstrator, is a powerful tool to assess the performance of various optical architectures for GEO and LEO optical links, which allows to point to optimal configurations for future operational OGS..

Laser-plasma accelerators emerge as ultra-compact and versatile sources for numerous applications. Although the acceleration length is short (a few millimeters), they typically require large-scale infrastructures including ultra-high-power lasers, vacuum chambers and strict stability for temperature and humidity. As a result, most experiments are conducted in laboratories in large areas with controlled environments. Here, we present a highly compact (footprint of ~ 9 m²) and transportable system capable of generating electrons and photons in the MeV range at high repetition rates (up to 10 Hz) with average charge levels of 0.5 nC and up to 1 nC. This achievement shows the feasibility of performing laser-plasma acceleration outside of laboratory environments with a transportable system, significantly expanding the potential for practical applications.

Many devices communicate externally in today's technologydriven world, creating cyber-attack vulnerabilities. To safeguard against these threats, utilizing an Intrusion Detection System (IDS) incorporating AI for effective anomaly detection is crucial. However, it is important to manage power consumption and latency carefully. Optimization of machine learning (ML) models can be achieved through three primary approaches: tuning hyperparameters and ML architecture via Neural Architecture Search (NAS) [1], enhancing existing ML architectures to improve robustness while reducing complexity [2], and optimizing for hardware implementation to strike a balance between Quality of Service (QoS) and Quality of Results (QoR) [3]. To address these challenges, we design a multi-level optimization approach that directly takes hardware and software constraints (such as QoS and QoR) into account during the optimization process with a hardware estimator (HE), greatly reducing the required number of actual ML architecture implementations. The main contributions are the fast exploration and evaluation of solutions based on a proposed hardware estimation regressor and the second is the ability to select and compare the most appropriate ML model architectures according to objectives and hardware constraints.

Eye-tracking research offers valuable insights into human gaze behavior by examining the neurophysiological mechanisms that govern eye movements and their dynamic interactions with external stimuli. This review explores the foundational principles of oculomotor control, emphasizing the neural subsystems responsible for gaze stabilization and orientation. Although controlled laboratory studies have significantly advanced our understanding of these mechanisms, their ecological validity remains a critical limitation. However, the emergence of mobile eye tracking technologies has enabled research in naturalistic environments, uncovering the intricate interplay between gaze behavior and inputs from the head, trunk, and sensory systems. Furthermore, rapid technological advancements have broadened the application of eye-tracking across neuroscience, psychology, and related disciplines, resulting in methodological fragmentation that complicates the integration of findings across fields. In response to these challenges, this review underscores the distinctions between head-restrained and naturalistic conditions, emphasizing the importance of bridging neurophysiological insights with experimental paradigms. By addressing these complexities, this work seeks to elucidate the diverse methodologies employed for recording eye movements, providing critical guidance to mitigate potential pitfalls in the selection and design of experimental paradigms.

In this paper, we present a realization of an on chip atomic clock using a cold cloud of $^{87}\mathrm{Rb}$ atoms. This clock is based on a Ramsey interferometer with a Ramsey time around 600 ms. This is realized with large laboratory temperature drift during the measurement (few degrees per day) and without magnetic-field shielding. We review the experimental implementation of this clock and give a full study of the known noises present in this atomic clock.

THALES Germany develops ion thrusters based on the HEMPT technology (High-Efficiency Multistage Plasma Thruster). The first fully qualified thruster model, the 1.4kW-class HEMPT-3050, has been integrated on the German geostationary satellite H2Sat, launched in July 2023, and is successfully operating since then with nominal performance. As a further development Thales Germany is qualifying a 700W-class HEMP thruster, called EVO. Thales Ulm has also started the development of a 1.5kW class thruster, the HEMPT 3050-C, both leveraging Thales in-orbit heritage and design for manufacturing approach. This paper will give an update on the status of the HEMPT product line at Thales, including recent results from development and qualification activities.

A spatial Hamiltonian model is built to describe the evolution of electrons and the electromagnetic field within a traveling wave tube (TWT), neglecting space charge and magnetic field effects. This model rigorously characterizes the wave-beam interaction and the associated momentum exchanges between the field and the electron beam, providing a detailed understanding of the amplification mechanisms in the TWT.