On-board data processing (OBDP) is gaining increasing attention in the industry. This is primarily because the diversity and volume of data generated by satellites are growing and business models are evolving to meet more demanding end-user requirements.
This article discusses on-board data processing with reference to SpaceCloud®, a payload computing hardware with a framework for satellite cloud computing applications created by the Swedish public company Unibap AB.
Unibap is a technology provider focussed on developing artificial intelligence (AI) and digitalization solutions for industrial and space applications, and is a paying participant of the satsearch membership program.
The problem: the growing data processing burden
Today’s satellites are far more than simple sensors in the sky. In recent years, innovations in Earth Observation (EO) and remote sensing, across a wide variety of spectra, have led to new satellite business models and data collection concepts.
For example, there are now optical payloads on the market with greater resolutions and range of functions in a small physical form factor. EO constellations based on such cameras have also been funded and launched, providing greater availability and reliability for end-users. In addition, there has been an increase in the number and diversity of those end-users, from both the public and private sectors, who have found new commercial opportunities based on the data available.
This growth has led to greater demands on the processing facilities of satellite systems in a number of key areas:
Storage constraints – as satellites collect more data (in terms of both the number of files and the average size of an individual file) the on-board storage capacity is under greater strain. Data downlinks are limited by the downlink capacity and ground station coverage, meaning the satellite itself can end up storing more data for longer periods of time, potentially affecting its ability to collect more. At the same time, the satellite itself is increasingly expected to utilize some of the data collected to make decisions on-board, meaning that the data’s accessibility, as well as the total capacity, is also under pressure.
The need for greater computer resources – more powerful and complex satellite applications are requiring greater computer resources throughout the system. Even the smallest satellites can now feature complicated sub-systems (such as deployable systems, adaptable propulsion units, and high-performance sensors) that all require control and coordination by the system’s On-Board Computer (OBC). This also places greater demands on system health monitoring and error reporting protocols.
The need for better in-orbit decision-making – there are growing demands on satellite manufacturers to produce systems that can operate in a more agile manner, to more easily position Earth Observation cameras for example, and provide greater versatility in missions. More agile and versatile satellites are being developed incorporating innovations in:
- Attitude determination (e.g. next-generation sensors),
- Attitude control (e.g. through the use of more advanced Attitude Determination and Control Systems (ADCSs), small Control Moment Gyroscopes (CMGs), and propulsion units), and
- Other forms of in-orbit transit and positioning technology.
However, aspects of such new capabilities can only work effectively if the satellite itself is capable of fast and effective decision-making, independent from ground control.
In addition, in-orbit decision-making can result in distributed decision-making, so that satellites can start reasoning with each other, do autonomous research, and control satellite constellations.
For example, this will be particularly important if a satellite is orbiting the Moon or Mars where the response time will be delayed by several minutes. Likewise, rovers exploring surfaces of planets looking for minerals, water, or indications of organic molecules would be able to more easily navigate and avoid getting stuck if they can operate autonomously.
Ultimately, the more advanced any space system becomes the greater the requirement for, and benefits that, increased autonomy can bring.
The downstream burden – limited downlinking bandwidth is a potentially significant constraint on an effective mission, particularly with the increase in the amount of sensor data collected and transmitted in today’s missions. In addition, typically, once data is downloaded to the ground station it is then corrected, calibrated, and possibly archived before being distributed to the various stakeholders that use the data to make decisions or create products. All of these processes need to be carried out efficiently and securely, and the cleaner and more well-organized the input data, the easier this is to achieve.
Such increases in demand are being looked at in a number of ways. In some orbits an optimally situated satellite could act as a data hub, for example; storing and downlinking data from several other satellites.
Optical inter-satellite communication links are also being established so systems can more easily transfer data, and there are also a wide variety of innovations in the ground segment to overcome potential bottlenecks in the system.
In addition, one of the key enablers of more efficient satellite services, and one that is gaining increasing attention in the market, is on-board data processing (OBDP).
The solution: on-board data processing
On Earth, data processing tools and capacity are ubiquitous. Access to computing resources with the ability to parse, compress, manage, analyze, and transfer image data (and other similar formats) is provided by a wide range of device- and cloud-based programs, many of which are free. However, in space the unique challenges of the environment dictate whether and how such tools can be used.
Many legacy satellites had very limited capacity for processing and handling data on board. Instead, they would simply store what raw data had been collected until it could be downlinked to the ground, in bulk or in packages.
This meant that unusable images, corrupted files, incorrectly formatted information, or other useless data took up valuable storage and downlink bandwidth, before being rejected on the ground once analyzed.
By processing the data on-board the satellite, some of these issues can be avoided. This makes the overall data collection more valuable and frees up more of the limited available communication bandwidth. This is particularly important for low latency operations and even for deep space missions.
Satellite OBDP enables the removal of unusable information, such as images outside the target area or that are obscured by clouds and smoke. This makes it faster and cheaper to download the valuable data to the ground, as there is less to download.
On-board processors can also compress data so that individual packages are smaller, again making them faster and cheaper to download.
For Earth Observation (EO) applications there is clear value in the ability to post-process captured images soon after they are acquired. This concept is sometimes described as ‘putting the brain closer to the eye’, or edge computing, and can be a core building block in the development of a more advanced satellite imaging service.
An advanced payload processor does, however, need to be carefully integrated into a satellite to ensure that it works efficiently with the payload and other on-board computer (OBC) technology.
Integration and operational challenges
Next-generation satellites already require greater computing resources to make full use of their tools and systems. At the same time, efforts are made to reduce the size and weight of satellites, requiring engineers to closely pack smaller components and sub-systems into the structure.
Payload processors then typically face the same environmental and operational challenges that can affect OBCs. COTS computation systems for space are often complex due to the stringent limitations on mass, power consumption, size, and timing and communication requirements. However, the most important factor is the required radiation tolerance.
As the size of processors increases, the risk of Single Event Upsets (SEUs) becomes more prominent. In addition, an increase in the clockspeed of the processor increases the number of latching windows per second during which the charged particles can impact data on the CPU memory.
In recent years there have been more rad-hard payload processors, OBCs, and electronic components brought to market that have opened up new opportunities for satellite integrators.
In addition, the growth of constellations means that operational models, and hence satellite integration requirements, are changing. OBDP capabilities on multiple systems in-orbit enable them to work in a similar manner to cloud computing on the ground.
To meet such requirements, satsearch member Unibap has developed an OBDP system utilizing cloud-based and cloud-inspired technologies.
Data processing in the cloud
Unibap is a Sweden-based computer system technology developer creating AI and computing technologies to support advanced automation systems for industries in space, as well as intelligent vision solutions to automate production flows on Earth.
The company has developed a system called SpaceCloud® – a payload computing hardware system with a software framework designed to facilitate timely, actionable distribution of information through on-orbit cloud edge computing, and data processing, storage, and analytics.
With SpaceCloud®, Unibap is aiming to accelerate the adoption of OBDP capabilities in the space industry, making them more accessible to teams building satellites around the world.
The SpaceCloud® software consists of both a framework and an operating system (OS) and has the following key features:
- A Linux Driver/API & Application Software Development – accessible tools designed to make the system more versatile and interoperable,
- A framework for utilizing and managing the resources available in SpaceCloud®
- Capabilities for AI algorithm development and implementation, designed to enable next-generation satellite applications,
- Software for effective data distribution, and
- A safety chip and safety boot to mitigate radiation effects.
More information on the SpaceCloud® suite of products is available on the pages below:
SpaceCloud provides a software framework for information storage and processing, utilizing flexible and reusable compute nodes in a distributed network in space as well as on ground.
The Unibap iX5100 is a SpaceCloud® computer solution featuring an embedded, x86-compatible, and AMD® G-series SOC product from the 1st, 2nd, & LX families. The SOC is paired with a powerful Microsemi® SmartFusion2™ FPGA, which provides IO extension and board supervisory management.
The next section discusses some of the potential applications that SpaceCloud® can enable in satellite missions.
On-board data processing in action
There are several existing and emerging space applications that are dependent on OBDP, such as:
- Edge cloud computing,
- Disaster monitoring,
- Cloud detection and removal,
- Ship or land-based vehicle detection,
- Space Situational Awareness (SSA),
- Synthetic Aperture Radar (SAR), and
- Autonomous vehicle operation.
The following sections present some of the in-orbit demonstration (IOD), research, and commercial missions on which the OBDP platform SpaceCloud® has been launched, tested, and verified.
D-Orbit’s Wild Ride ION mission
The Wild Ride ION mission was launched on the 30th of June 2021 on a SpaceX Falcon 9 rocket from Cape Canaveral, Florida. The in-orbit validation mission included a variety of applications targeting some of the highest-value space data market segments, such as;
- Disaster monitoring,
- Video and image data transfer optimization,
- Space domain awareness (SDA),
- Advanced image processing for precision agriculture,
- Defense early warning, and
- Integrated satellite communication.
To illustrate some of these uses further, below we look at specific mission examples in which SpaceCloud’s OBDP capabilities have been deployed:
D-Orbit’s ION Satellite Carrier mission
Firstly, again launching from Cape Canaveral, Florida, this time onboard SpaceX’s Transporter-3 mission, this mission features 17 SpaceCloud applications and is designed to further demonstrate how a satellite can become a more versatile, configurable, and multi-purpose instrument.
It also features the ability to upload new applications while in-orbit so that testing regimes can evolve, and new ideas may be brought on-orbit faster. In addition, the system also includes a hyperspectral electro-optical instrument developed by VTT that is giving end-users access to almost real-time Earth Observation imagery.
Working with the European Space Agency (ESA)
Unibap has an ongoing agreement with ESA to demonstrate the capabilities of SpaceCloud® in-orbit. The objective is to investigate aspects of the software-defined satellite concept – working with the existing hardware and software on an orbiting system and changing operational aspects to meet new goals.
In December 2021, Unibap successfully demonstrated the ability to reconfigure a sensor on an operational satellite in order to meet new mission criteria. An existing star tracker was reconfigured to capture new EO data, masking clouds and unnecessary land areas.
In just 3 weeks a solution was designed, the satellite’s neural network retrained with new input uploaded from the ground, operating algorithms changed, and new data was captured and downlinked. This mission demonstrated that more versatile satellites are possible with today’s hardware and sub-system setups.
The NASA Distributed Spacecraft Autonomy (DSA) research
NASA selected Unibap to provide 25 SpaceCloud® products from the company’s iX5-family for the DSA research program.
The Unibap products are deployed in ground-based research at the NASA Ames Research Center, with the aim of advancing autonomous capabilities for distributed space systems.
Delivery was made during the first half of 2021 and Unibap is supporting NASA with on-site integration, in partnership with Unibap’s US distributor.
The Hyperspectral Thermal Imager (HyTI) mission
The iX5-100 computer has also been used in the Hyperspectral Thermal Imager (HyTI) mission, funded by NASA’s Earth Science Technology Office InVEST, investigating volcanic degassing, land surface temperature changes, and precision agriculture metrics.
The system provides health monitoring, data transfer, processing, and storage functionality and features a heterogeneous architecture (CPU, GPU, FPGA). It also performs as an alternative dedicated AI accelerator.
Near real-time EO and mid-air airplane detection
The L3Harris™ geospatial intelligence software system ENVI®/IDL® was integrated into SpaceCloud® within a duration of 4 hours to deploy a near real-time Earth Observation application.
The application also featured machine learning capabilities developed by US-based remote sensing company SarianaSat™ Inc.
Another SpaceCloud® application developed by SaraniaSat™ Inc for US Space Force also leveraged the onboard ENVI®/IDL® L3Harris geospatial software suite for mid-air airplane detection.
The application scans 100 km2 of World View-3 MSI spectral data to produce geolocated coordinates for detected aircraft in under 1 minute on the iX5-100 computer.
The data processing burden for satellites is expected to further increase in coming years, as are customer and commercial demands for more agile satellites and systems able to adapt to changing mission criteria.
On-board data processing has the potential to improve many different satellite applications and generate greater value for their operators and downstream customers.
Implementing cloud-based systems that are capable of increasing versatility and operational performance is more of a question of development workflow and changing traditional procedures than it is of technology.
Solutions, such as Unibap’s SpaceCloud®, are already available on the market and are being tested and qualified in a wider range of application areas with each new mission.
Satellite operators with relevant business cases now have greater access to versatile OBDP systems that can more easily integrated into systems in development and that can bring benefits across mission lifetimes.
Unibap is investing heavily into SpaceCloud® and its ecosystem because they aim to propel this new edge computing market in space. SpaceCloud® is making it very simple to deploy new applications and tailor satellite missions, even down to changing mission requirements in-orbit, every second.
To find out more about Unibap, please feel free to view the following resources:
- Unibap’s satsearch supplier hub
- Footage from our webinar, a guide to advanced data processing and AI for satellite missions, in which Mathias Persson, Senior Business Development Manager at Unibap presented
- A podcast on implementing onboard AI processing capabilities for Earth Observation (EO) missions with Søren Pedersen, SpaceCloud Sales Manager at Unibap and Thys Cronje, CEO of Earth Observation payload manufacturer Simera Sense
- A podcast with Fredrik Bruhn from Unibap on Artificial Intelligence (AI) and edge computing in space