The Software-Defined Future of Satellites

Unstable operators’ business cases together with advances in digital technology impel satellite manufacturers to provide software-defined solutions that would be cheaper and more flexible than traditional satellites.

In the next months, American satellite manufacturer Lockheed Martin will launch a bunch of cubesats into the Low-Earth Orbit (LEO). These cubesats, however, will be somewhat different from regular run-of-the-mill cubesats. They will attempt to fly in formations, act like a space-based cloud computing platform, process data on board, and have their functionality changed through software updates beamed from the ground during the mission. The demonstration will present Lockheed Martin’s entrée into the era of software-defined satellites, which rely on rather generic hardware but flexible software to define their missions.

Lockheed’s contribution to the nascent technology is called SmartSat. For the upcoming experiments it will be combined with the company’s new SpaceCloud and HiveStar technologies, the latter being responsible for the swarming behavior.

According to Adam Johnson, SmartSat program manager at Lockheed Martin, SmartSat is essentially an operating system, or rather an operating environment, something like iOS for satellites. In the future, the system could run on all types of Lockheed Martin’s spacecraft from the smallest cubesats to the flagship geostationary platforms. The operators could then upload whatever applications they need based on the requirements of their missions.

“We sell a satellite to the customer and then they will have the ability to have multiple applications on board of that satellite,” Johnson says. “It’s more like a smartphone and everything is an application and you can start and stop those applications as much as you do it on your smartphone. It’s very convenient to reconfigure the vehicle.”

The Need for Flexibility

Flexibility and the ability to reconfigure a satellite which is already in orbit, is something operators have been asking for years. Especially for operators of Geostationary Orbit (GEO) satellites, which usually have lifespans of 15 years or more, the ability to adjust the spacecraft to the changing needs of the market is essential. That might include moving a satellite into a different position or even switching its functionality from TV broadcasting to internet connectivity — something that would be impossible with traditional hardware-defined satellites.

“Operators no longer have stable business cases for 15 years,” Airbus Head of Space Systems Jean-Marc Nasr said at the World Satellite Business Week in Paris in early September. “It’s more like five years now. And they are telling us that they need to be able to change the mission and that it needs to be cheap.”

Airbus has introduced its fully reconfigurable software-defined geostationary platform OneSat earlier this year and had already won a contract with Inmarsat to manufacture three satellites for the London-based operator’s next-generation Global Xpress Flex system.

The three Ka-band satellites, GX7, 8 & 9, to be launched after 2023, feature on-board processing and active antennas, and will be able to adjust their coverage, capacity and frequency.

Airbus said the OneSat platform, based on a standardized, modular and design-to-manufacture approach, could be delivered more quickly than existing telecommunications satellites.

At the World Satellite Business Week, Boeing and Thales Alenia Space both responded to Airbus’ offering with their own software-defined platforms.

Boeing’s 702X family of software-defined satellites includes a 1,900 kilogram (kg) GEO product, as well as a smaller Medium-Earth Orbit (MEO) platform. Boeing said the 702X could be delivered to customers within three years and would allow operators to distribute capacity to a variety of end users.

Thales Alenia Space introduced its Space Inspired (for INstant SPace In-orbit REconfiguration) satellite product line, which will enable mission and service reconfiguration, instant in-orbit adjustment to the demand, and a transition from video broadcasting to broadband connectivity services. Thales said the all-electric satellite would be low cost thanks to serial production.

“We very often build satellites one by one. It’s like craftsmanship, so higher cost comes with that,” says Didier Leboulch, Thales Alenia Space’s head of strategy and telecom solutions. “One of the benefits of the software-defined approach is to standardize the product, either at the block level or at the full satellite level. And add the specification through the software. We can expect to have a lower cost of production and to have our customers benefit from this lower cost.”


According to Shagun Sachdeva, a space industry analyst at Northern Sky Research (NSR), the jury is still out on how much software-defined satellites will shape the future.

“Software defined satellites will definitely hold a big part of the future but they are not going to be all of it,” Sachdeva says. “I believe that software-defined technology would be particularly interesting for the GEO operators, who are currently facing a lot of uncertainty and would benefit from software-defined technology, which not only provides the flexibility they need but could also reduce cost in the future.”

Geostationary satellites usually have the longest lifespans and the highest price tag, which makes them the most vulnerable to the instabilities of the market.

However, manufacturers and operators have already introduced partially software-defined satellites for LEO and MEO constellations as well.

In February this year, Iridium completed the deployment of its Next constellation of 75 LEO satellites, which had been manufactured by Thales Alenia Space. The narrowband constellation, described as the most sophisticated telecommunications system in the world, provides voice and data services to users in remote locations all over the planet.

“Iridium Next satellites have a processor onboard with software that you can reprogram,” says Thales’ Leboulch. “We can upgrade the software in order to deliver new, improved services that the old satellite could not provide.”

He adds that in narrowband satellites, such as those of the Iridium constellation, the software-defined approach is easier to implement compared to broadband geostationary platforms.

“The more bandwidth, the more throughput you are processing, the bigger the challenge,” he says. “Nevertheless, it is achievable today with the progress of digital technologies.”

Luxembourg-based satellite operator SES has been among the earliest advocates of the software-defined technology. The company’s upcoming MEO constellation O3b mPOWER, which is set to enhance the existing O3b, will be based on a partially software-defined approach, which will allow unprecedented flexibility in terms of bandwidth allocation.

“O3b mPOWER is not in the strictest sense of the term software-defined but has definable capabilities,” says Stewart Sanders, executive vice president of technology, SES Networks. “We now have a fully digitized payload as well as full electronic beam forming.”

The satellites are, however, not defined only through software but run on bespoke chipsets, which are unique for each mission. The fully software-defined approach, Sanders said, would at this stage be rather inefficient. Still, the advantages are substantial.

“We have reduced the number of analog components, which obviously reduces the size of the satellite, its mass and the cost,” Sanders says. “The biggest advantage, however, is that we can move into complete digitalization of the spectrum and that gives us an enormous amount of flexibility in what we can do with that spectrum.”

The channels can be defined flexibly based on the actual needs of customers in any given moment and adjusted based on the changing demand.

“When you couple that with electronic beam steering, which is what we are doing with O3b mPOWER, we can generate a beam specifically for each customer and give them exactly the amount of bandwidth they need.”

Traditionally, customers would buy a fixed amount of bandwidth, usually higher than what they actually need, Sanders said, which would mean that a lot of capacity would go unused. The flexible approach enables the satellite operator to use available bandwidth more efficiently and potentially serve more customers. The customers, on the other hand, only pay for what they actually use.

Maturing Technology

Still, technological hurdles remain to be solved before fully software-defined satellites can truly take off.

“Amplification is one of the bottlenecks for these software-defined satellites,” says Leboulch. “Software does not amplify the signal so at some point you need to go out of the software and the computer to amplify the signal to go down to Earth.”

Solid-state power amplifiers that are being used in active antennas on board of software-defined satellites are still less efficient than traditional technologies, says Leboulch.

“Due to the technological inefficiency of the solid-state power amplifiers, if you just plug a software-defined payload into a normal satellite, you would have less power on it,” he says. “That’s why, usually, if you want to have efficient software-defined satellites, you don’t need to change just the payload, you need to change the full platform to deliver more watts and more dissipation capability to your payload in order to equal the performance of conventional satellites.”

According to Lockheed Martin’s Johnson, new technology that would fully unleash the potential of software-defined satellites is just reaching maturity.

“The availability of the multicore processors is the first step and that’s the step we are taking right now,” he says. “A regular desktop computer on the ground could have something like eight cores or even more. But we have only just recently started to get multiprocessors from our suppliers that are radiation tolerant and radiation hardened and can survive in space.”

Further advancements in technology will allow capabilities to gradually improve and eventually power sophisticated AI algorithms.

“Hardware that allows greater processing on orbit for AI machine learning will become very important,” says Johnson. “GPUs are heavily used for AI machine learning on the ground and eventually having that capability in space could unleash a whole new world of opportunities.”

Intelligent Ground Segment

The innovation doesn’t stop with the satellite. Thales Alenia Space CEO Jean-Loic Galle said the company is looking to develop digital ground infrastructure that would be able to automatically interface with the digital assets in space and manage the digital payload.

SES is moving in the same direction and has recently announced its partnership with Kythera Space Solution to develop a software system called ARC (Adaptive Resource Control) that would dynamically synchronize the space and ground-based assets.

“The new payloads represent a big step in capability but also a big step in complexity,” says Sanders. “It’s no longer realistic to manage that complexity without software components on the ground as well.”

More automation of the overall ecosystem is what the players expect going forward. Further changes will come with the development of on-board data processing and satellite crosslinks, which would enable individual satellites to communicate with each other, says Johnson.

“The advent of terrestrial cloud computing enabled many revolutionary applications, such as Uber, for example,” said Johnson. “At that time, no one knew what that would do and we are at a comparable moment in space. It’s much more difficult to create the cloud environment in space and we are only just starting, so ten years from now, we will see what will come out of it.”

Astranis: Leveraging Software-Defined radio for cheap GEO coverage

San Francisco-based start-up Astranis believes that using a software-defined radio on board of its satellites would enable economies of scale in the manufacturing process and thus considerable cost reduction.

“Each satellite will essentially be identical to the other satellites,” says Astranis CEO John Gedmark. “The payload of the satellite can be configured very late in the production process or maybe even once it’s already on orbit.”

Astranis hopes to build a constellation of 350 kg geostationary satellites that would provide patches of connectivity to areas in need. Each of these satellites would have just a fraction of the transponders typical for a regular geostationary satellite and have a much smaller footprint.

“You can think about it as disaggregation,” says Gedmark. “We are taking the capacity or the capability of a very large traditional GEO satellite and essentially breaking that up to smaller chunks and deploying one chunk at a time where it is most needed rather than doing it all at once.”

Gedmark said that analogue repeaters that are traditionally used on board of GEO satellites greatly limit what can be done with the payload. The software-defined radio enables digital signal processing, which allows the operator to adjust frequencies, coverage and bandwidth based on the actual needs of the customer.

“You can even adjust the waveforms that you are supporting as industry standards evolve,” he says. “When there is a new waveform that customers want to use, you can apply this new wave form. There is a whole range of things that you can do.”

The company has flown a prototype, called DemoSat-2, in 2018, and successfully tested the software-defined radio to uplink HD videos to the spacecraft, process the signal in real time and downlink it to a ground station in Alaska.

The team is now working on its first commercial mission, which promises to triple the satellite internet capacity available to the U.S. northernmost state.

“We will have about 7.5 gigabits per second of capacity,” Gedmark says. “That’s a lot of bandwidth, so the digital processing power in the software-defined radio has to be able to handle all of that simultaneously. The second challenge is qualifying the electronics to survive in the radiation environments of space.”

Astranis expects to refresh its assets much more frequently than regular GEO operators and only plans for a seven-year lifespan of its satellites. Gedmark believes there is a need for this type of technology to complement the existing geostationary coverage and expects the firm to launch dozens of satellites in the coming years. VS

previousAdditive Manufacturing Will (and Should) Change EverythingnextRegional Roundup: June 2015