Satellite Signal Processing Techniques using a Commercial Off-The-Shelf AI Chipset (SPAICE)

SPAICE develops and validates AI-powered onboard signal processing for satellite communications using commercial off-the-shelf (COTS) AI chipsets. By bringing intelligence to the network edge, the project enables real-time decision-making, reduces latency and ground dependence, and advances autonomous, adaptive satellite communications for future non-terrestrial networks.

The project at a glance

Start date:

11 Nov 2021
Duration in months:

55
Funding:

European Space Agency
Principal Investigator(s):

Symeon CHATZINOTAS

��Vlog

The Satellite Signal Processing Techniques using a Commercial Off-The-Shelf AI Chipset (SPAICE) project develops and validates artificial intelligence (AI)-based signal processing techniques for satellite communications, such as signal identification, spectrum monitoring, spectrum sharing and demodulation, using a commercial off-the-shelf (COTS) AI chipset under ESA Contract No. 4000134522/21/NL/FGL. The primary objective is to leverage onboard intelligence to enhance decision-making in satellite communication networks, particularly for adaptive resource management in radio access networks (RANs). By processing data directly on edge devices, these machine learning (ML) models reduce latency, minimise ground-station dependency and improve satellite autonomy. To demonstrate these advances, a laboratory testbed was developed to validate performance in both ground-based and onboard scenarios. The targeted improvement over the state of the art is to enable real-time onboard satellite signal processing that would otherwise be impossible to execute without dedicated AI accelerators. The findings provide valuable insights into the feasibility of integrating onboard, AI-driven decision-making into next-generation space systems. The proposed methodology contributes to the growing body of work on embedded ML for space applications, paving the way for more autonomous and intelligent satellite networks. By addressing the fundamental challenges of deploying ML in space environments, this work supports future advances in edge AI, intelligent communications and adaptive resource management for non-terrestrial networks (NTN).