07-12, 14:00–14:30 (Europe/Amsterdam), For Loop (3.2)
The deployment of low Earth orbit satellite internet constellations has revolutionized global connectivity. The design and operations of these systems rely heavily on large-scale simulations. At Eutelsat-Oneweb, we have developed a comprehensive full-scale constellation simulator to power our design and regulatory compliance workflows. This talk provides an overview of the simulation architecture and performance benefits that the use of Julia has unlocked for us.
Satellite internet constellations are global scale systems designed to deliver ubiquitous low-latency and high bandwidth connectivity across the globe. In these systems users are issued terminals to communicate with the constellation which relays information to and from gateways on the ground. User traffic is then routed into other networks. Notable examples of these systems include the existing Starlink and OneWeb constellations, with ongoing efforts in deploying similar systems.
In the context of Eutelsat-OneWeb's system, thousands of subsystems are involved in delivering our connectivity services. The orchestration and intricate coordination amongst satellites, user terminals, semi-automated control centres, expansive internet networks and ground antenna arrays is critical for ensuring seamless service.
A key aspect of the design and operation of such constellations is modelling the interaction between subsystems. For example, to estimate the impact of system configuration changes on the quality of service. Simulations become particularly detailed in that case, necessitating second level resolution over extended durations.
To support the design of our next generation constellation, we undertook a rework of our simulation tooling. The guiding objectives were to enhance the flexibility and performance to facilitate rapid design iterations.
This presentation is an overview of our implementation of a simulator designed to meet our design goals with minimal complexity. Specifically, we focused on the following areas:
- Simulating of all our subsystems and their variants in one unified environment.
-
Enabling the configuration and development of the simulator without imposing stringent programming knowledge requirements.
-
Achieving high performance to enable fast iteration on simulation results.
We'll explain how we use Julia's unique features to implement a comprehensive system simulator supporting arbitrary subsystem models transparently. In particular, how this approach allows us to develop simulations efficiently.
Additionally, we will also describe the performance and large-scale deployment benefits we encountered through the switch to this language.
Romain develops satellite constellations as part of the systems engineering team of Eutelsat Group. Having studied aeronautics and mathematics, his prior experience is mainly in unmanned aircraft design.