ESA NeuroSat: “The Application of Neuromorphic Processors to Satcom Applications"
Project Description:
Mimicking the way a human brain communicates, bioinspired electronics based on non-volatile memory devices have emerged as a promising alternative to traditional computing technologies based on transistors. While conventional processors operate on a batch mode, i.e. batching up large number of samples before processing them, the new neuromorphic processors (NPs) can process data samples as quickly as possible.
One of the main driving forces for investments in neuromorphic computing is the promise it holds for Artificial Intelligence (AI). While conventional computers are still preferred for specific applications, NPs are ideally suited for tasks that human-brain can solve easily, such as recognition and reasoning. Currently, neuromorphic computer systems are still being researched. So far, there is still a lot of development and prototyping going on. The technology is quickly gaining momentum and large corporations, such as Intel and IBM, are participating in research projects related to this promising technology.
NPs represent a major opportunity to unlock the potential benefits of AI and ML solutions for satellite communications (SatCom) systems thanks to their energy efficiency and capability of continual, on-board, adaptation.
The project NeuroSat aims to identify the potential benefits of using NPs implementing spiking neural networks (SNN) for SatCom applications. The study's primary outcomes will be to inform the community of the capabilities and feasibility of implementing neuromorphic processors in SatCom systems and to define the technological developments needed to make them a reality.
The proposed activity takes the form of an investigation into the implementation of NP architectures in SatCom systems, focusing on AI-based applications that require massive parallelism, processing of streaming data, energy efficiency, and adaptation. We identify such applications, describe NP systems tailored for SatCom, present a prototypical implementation using Intel’s Loihi NP, and trace a path ahead for research and implementation of neuromorphic computing for SatCom.
The project was carried out by University of Luxembourg (UL), which contributes its expertise on SatCom, and King’s College London (KCL), whose team members have a longstanding experience in the design and implementation of NPs and SNNs.
Funding: This project is funded by the European Space Agency (ESA) ARTES Future Preparation (web from ESA)
Project Duration: 2022-2023
Relevant Publications:
F. Ortiz, N. Skatchkovsky, E. Lagunas, W. Martins, G. Eappen, S. Daoud, O. Simeone, B. Rajendran, S. Chatzinotas, “Energy-Efficient On-Board Radio Resource Management for Satellite Communications via Neuromorphic Computing”, submitted to IEEE Trans. On Machine Learning in Communications and Networking.
NeuroSat Workshop 12 July 2022
Research Team:
Dr Eva Lagunas (SnT) – PI
Dr Flor G. Ortiz (SnT) - VPI
Dr. Wallace Martins (SnT)
Dr. Geoffrey Eappen (SnT)
Dr. Saed Daoud (SnT)
Dr. Jorge Querol (SnT)
Dr. Jan Thoemel (SnT)
Prof. Symeon Chatzinotas (SnT)
Prof. Osvaldo Simeone (KCL)
Prof. Bipin Rajendran (KCL)
Dr. Nicolas Skatchkovsky (KCL)