Constellations spoke with Sethu Saveda Suvanam, Founder and CEO of ReOrbit, about the digital transformation of the satellite industry. With the increasing prevalence of software-defined satellites and increased capacity for onboard computing, the landscape of the entire sector is changing quickly.

The Rise of the Software-Defined Satellite

In the majority of currently flying satellites, data moves in point-to-point trajectories. That’s vastly different from the complex, distributed network architecture that exists on the ground. “We are connected over a digital world where the data is flowing very seamlessly on computers, on mobile networks,” Suvanam says of ground networks.

Software-defined satellites are changing the game. By making it possible for satellite constellations to have similar complex data relationships. “[We] can start designing a satellite like a flying router,” Suvanam says. “You could have multiple sources of data coming into your satellite, and also multiple sinks of data from your satellite.”

This development opens the “Pandora’s box of applications,” Suvanam says. A software-defined satellite “can now start connecting with all the satellites that are around, within the same orbit or multi-orbit.” This transformation is similar to changes in other industries that have taken place over the last few decades. “If you look at industry verticals that are very similar to satellite, like computers, automobiles and cell phones, all these industries have understood probably 30, 40 years ago that the core is actually software intelligence.”

Data Movement and Onboard Processing

One of the advantages of a software-defined satellite is the capability for onboard processing, which Suvanam describes as existing in two primary layers. “Basically, you should look at it in multiple layers, where one, it is computing and distributing data itself. And two, utilizing, basically, the unused capacity of different satellites as a whole to do advanced processing.”

In the first layer, the satellite should be able to identify and categorize data. For example, a satellite that primarily collects Earth observation data may be connected to a telecom constellation—two different applications with different kinds of data. A software-defined satellite “should be able to process all these different kinds of data coming in,” Suvanam says. In the second layer, free-flowing data allows satellites to tap into unused processing within a satellite network. Instead of sending data to the ground to be processed, the data can be processed in space with onboard computing.

Artificial intelligence is also an important benefit of advanced onboard processing. “When we talk about AI and machine learning, we are basically trying to bring all that magic that’s happening on the ground today onboard the platform,” Suvanam says. By processing raw data into useful intelligence, “we could capture a picture, we could process all that data, we could build useful information, and send it directly to the end user on an app.” And all of this in a matter of minutes or seconds.

Looking Forward: Cost and Timeline

“Software-defined spacecraft are typically expensive as compared to the traditional payloads,” Suvanam says. If you don’t need advanced high-data processing, you might not have a pressing need for software-defined satellites. But that comes with a caveat—Suvanam believes that “within 10 to 15 years, every satellite will be a software-defined satellite. I cannot see a use case for traditional satellites in the next 15 years.”

“It’s just a matter of time,” Suvanam says, noting the increasing use cases for software-defined satellites and their availability to larger masses. “We are at the stage where we could compare it to the 80s or 90s in the computer industry, where the computers have become commoditized.”

For more on satellite use cases, machine learning and software flexibility, listen to the full episode here.