AUTSS

Competitiveness & Growth

Status

Completed
Status date

2024-05-10
Activity Code

6B.080

Objectives

The AUTSS platform was completed in 2021 as an incubator and accelerator for AI technologies in the Space industry. Currently AUTSS is being used to develop the Dynamic Predictive Routing and Receiver Chain Optimisation use cases with ARTES.

Dynamic Predictive Routing

Where satellite communications missions operate in LEO or MEO orbits, the communications links change frequently due to the relative motion of spacecraft above the Earth and with respect to each other.
The Dynamic Predictive Routing (DPR) project will leverage the AUTSS AI accelerator to determine if and how Artificial Intelligence and Machine Learning can be applied to optimise communications paths, improve traffic routing and make better use of bandwidth and increasing operational efficiency.

Receiver Chain Optimisation

The European market for lunar communications is poised for growth, driven by increasing interest and investment in lunar exploration from NASA’s Artemis programme coinciding with the first generation of commercial lunar spaceflights beginning with Intuitive Machine’s landing on the moon in February 2024.
Goal 1: Reducing the capital expenditure by moving to a software-based receiver chain, creating a virtualised receiver chain that can be duplicated as required.
Goal 2: Increase antenna performance to an equivalent larger size reducing the size for the antenna required to support Lunar communications from 16-18m to <14m. This will allow the use of existing smaller and cheaper antennas and mitigate the need to develop a larger class of lunar antenna in the 16-18m range.
Goal 3: Enable more data to be received in the same amount of time giving “better value for money” per pass. Better efficiency will reduce the time and financial costs of repeat passes and free up ground station time which can be directed to other services.
Goal4: Maximise the productivity of spacecraft science output with more data available in the same duration. This will enable more research, publications, and more robust outputs from a short mission duration increasing the commercial investment case for short duration lunar exploration.

Challenges

Machine learning technologies are cutting-edge technologies, their applications and behaviours are not always well understood or documented.

Capabilities others might consider to be ‘the hard stuff’ (big data, DevOps, automation) are just the simple prerequisites for leveraging AI technologies.

Benefits

AUTSS provides intuitive and de-mystified access to machine learning technologies for small and medium size enterprises in the Space industry.

Partner organisations can quickly and effectively assess the viability of leveraging AI for a particular use case and, where viable, to seamlessly industrialise a prototype as a delivered product using Machine-Learning-as-a-Service (MLaaS) methodology.

Features

AUTSS offers an on-premises computing platform dedicated to data science activities with an industry-leading security design and strong operational controls from inception.

The AUTSS platform is portable can be deployed to the cloud or on your own premise to suit all business needs.

Leveraging CGI’s long history in the Space and Satcoms industries allows AUTSS to provide industry-specific tooling, assets and knowledge that is not available from large generic vendors. Bespoke machine-learning tooling is designed from the ground up for the challenges of the space industry.

System Architecture

The AUTSS platform is a bespoke data science environment built on powerful NVidia Tesla GPUs in our secure facility in Leatherhead, UK. AUTSS draws on a co-located data lake of reusable open-source assets with secure and compartmentalised storage for sensitive data.

AUTSS has the capacity to access cloud services and is fully portable depending on customer needs; it can interface directly to partner organisations to access data pipelines where necessary. These external surfaces are protected by CGI’s SecOps Cyber Security service.

The AUTSS platform itself is built on MLaaS principles and has the capability to support model training from conception through industrialisation and operationalisation into service use.

AUTSS also offers portability features to allow a partner organisation to integrate a machine learning model in-house with minimum fuss after initial prototyping has completed successfully.

Plan

Dynamic Predictive Routing initiated its Definition Phase in May 2022 and will conclude its Technology Phase in June 2024.

Receiver Chain Optimisation has begun a Product phase in February 2024 and will run until Transfer to Operations by May 2025

Current status

AUTSS accelerator platform has been developed and is in use supporting several industry engagements.

For ARTES, AUTSS is being used to investigate the Dynamic Predictive Routing and Receiver Chain Optimisation use cases.