Synthetic data generators (SDGs) are indispensable tools in telecommunications and satellite communications, providing a means to simulate otherwise hard-to-obtain realistic traffic scenarios for pre-deployment testing and system optimization. This paper introduces a framework for generating synthetic bandwidth demand data by integrating macro-scale and micro-scale approaches. The macro-scale SDG models long-term trends, daily and weekly seasonality, random noise, and occasional spikes over an extended period of time, typically a year. In contrast, the micro-scale SDG captures short-term, minute-level variations within a day or week, adjusted for different application demands such as phone calls, video calls, and video streams. The proposed ensemble SDG merges these scales, producing synthetic datasets that provide high fidelity in both broad and granular temporal views of bandwidth demand. We further extend the model by scaling demand with population density and projecting it onto satellite beam footprints for SatCom applications. This paper details the mathematical formulations, implementations, and theoretical underpinnings of each SDG component, demonstrating their effectiveness and realism through experimentation. The proposed framework supports a wide range of applications, enhancing the ability to plan, optimize, and innovate in the field of (not only satellite) telecommunications.

The increasing deployment of autonomous machine learning systems on-board spacecraft is vital for advancing space exploration and operations, yet ensuring their reliability presents significant Verification and Validation (V&V) challenges due to resource constrained compute and communication limitations, the overall harsh space environment, and the increased need for dependability involved in the domain. This paper addresses the critical need for robust V&V methodologies tailored to these demanding conditions. We outline a systematic mapping study linking space applications with pertinent ML areas and architectures, alongside a comprehensive review of suitable design-time and runtime verification techniques addressing unique spacespecific factors. Central to this is a novel proposed framework and associated dual Model/System pipeline architecture based on MLOps principles, facilitating rigorous V&V across the ML lifecycle. By providing this structured approach, illustrated with relevant use cases, we aim to enhance trustworthiness and support the qualification of autonomous ML for dependable onboard ML deployments in space missions. Index Terms—Autonomous Systems, On-board machine learning, Space applications, Formal Verification

This paper presents a Satellite-As-a-Service (SaaS) architecture designed to enable flexible and efficient deployment of Machine Learning (ML) workloads on heterogeneous edge hardware platforms in space. Leveraging container-based virtualization (Docker) and an orchestration framework (Kubernetes), our approach abstracts hardware complexity and supports a variety of accelerators — FPGAs, TPUs, VPUs and NPUs —within a unified development and deployment environment. We integrate DevOps design principles delivering a reconfigurable stack that supports rapid ML model updates and deployment on target hardware. By treating satellites as extensible service platforms, we demonstrate how containerization and hardware abstraction streamline the onboarding of advanced ML algorithms, ranging from convolutional neural networks for image processing to neuromorphic paradigms for ultra-low-power inference. We detail how standardized APIs and modular workflows promote interoperability across multiple satellite systems and heterogeneous hardware accelerators. Overall, the presented SaaS architecture offers a pathway toward smarter, more versatile satellite payload operations, shaping the next generation of inorbit data processing and autonomy.

Satellite Communications (SatCom) are a backbone of worldwide development. In contrast with the past, when the GEO satellites were the only means for such connectivity, nowadays the multi-orbital connectivity is emerging, especially with the use of satellite constellations. Simultaneously, SatCom enabled the so-called In-Flight Connectivity, while with the advent of 5G-NTN, the devel opment of this market is being accelerated. However, there are still various missing points before such a technology becomes mainstream, especially in the case of Rotary Wing Aircraft (RWA). Indeed, due to their particular characteristics, such as the low altitude flights and the blade interference, there are still open challenges. In this work, an End-to-End (E2E) analysis for the performance of SatCom under 5G-NTN for manned and unmanned RWA is performed. Various scenarios are examined, and related requirements are shown. The effects of blades and other characteristics of the RWA are established, and simulations for these cases are developed. Results along with related discussion are presented, while future directions for development are suggested. This work is part of the ESA ACROSS-AIR project.

Extreme-edge computers coupled with sophisticated remote sensors are becoming a pivotal component for Earth Observation (EO) satellites. Such components benefit from the employment of virtualization techniques to enable the dynamic deployment of novel on-board Artificial Intelligence (AI) and to concurrently support multiple users on the same platform. This paper presents a comparative analysis of state-of-the-art virtualization techniques (namely Unikernels, Virtual Machines and Containers) specifically within the realm of AI-driven EO. Focused on five key criteria – security, scalability, resource efficiency, performance, and multitenancy – the study synthesizes existing literature to elucidate the strengths and limitations of each virtualization technology, and augments this understanding through a hands-on evaluation. Unikernels are distinguished for their minimalistic design and high efficiency, virtual machines for their robust isolation and stability, and ontainers for their flexibility. The comparative framework aims to guide engineers in selecting the most suitable virtualization technology according to the needs of their use case. This analysis clarifies the current state of virtualization technologies and provides a nuanced understanding of their applicability in advancing the capabilities of AI-driven EO systems.

Given a mission profile, using very similar components but different configurations can allow an Earth Observation satellite to acquire large swath observations, offering very high revisit frequencies, at the expense of lower spatial resolution or vice versa. In the feasibility study summarized herein, the objective is to have a wide-swath optical camera for the maximization of the coverage, combined with a high-resolution, narrow-swath camera to achieve the highest required spatial resolution and get the best possible image of an Area-of-Interest (AoI), with the help of on-board AI decision making. The performed analysis concludes with a case study demonstration on active fire detection.