Constellations interviewed a panel of industry experts, including Vinay Purohit of CloudJuncxion, Dallas Kasaboski of Analysys Mason Space and Kevin Tobias of Kratos, about digitizing the edge of satellite networks.

Constellations interviewed a panel of industry experts, including Vinay Purohit of CloudJuncxion, Dallas Kasaboski of Analysys Mason Space and Kevin Tobias of Kratos, about digitizing the edge of satellite networks. The discussion included the movement toward digital architectures, redesigning the network edge and the use cases for AI in software-based satellites.

All About the Edge

The “definition of the edge is highly dynamic,” said Purohit, Founder and CTO of CloudJuncxion. “If you’re in a first responder network, edge is where the first responders are. And that may change if you are in the oil and gas industry, that may be out in the ocean.”

Oftentimes, the edge of a satellite network refers to the user terminal. And the edge device or user terminal market is also changing rapidly. “Five years ago, the terminal market was limited. There were a lot of interested players, a lot of development, but scale was low,” said Kasaboski, Principal Analyst at Analysys Mason Space. “Now, [there’s] greater pressure, more players, more investment and we’re seeing a lot of real, viable products on the floor.”

Panelists noted the growth of this market is largely related to digitization, which is being fueled by two major drivers: standardization and flexibility.

More companies are looking for edge devices that enable scale, growth and interoperability. But moving to digital or software-defined products is not an easy switch to make. “You have to think about the full life cycle, installation, operations and maintenance,” said Tobias, Director of Product Management at Kratos. “The first thing we do in provisioning the terminal is to bring those components together based upon the use case.”

Multi-Orbit, Multi-Beam and Multi-Everything

One of the first steps in digitizing the network edge is disaggregating software and hardware. “We’ve disaggregated the network function from the proprietary hardware,” said Tobias. “It being software-defined and virtualized is basically the basis for making that happen. It’s table stakes.”

“Once we’re running the network functions at the edge, we can orchestrate those and run them on a single platform. It also serves as a basis for thinking about the entire edge, or the entire service, that the operator is trying to provide,” said Tobias. This also opens up opportunities for more consistent multi-orbit capabilities.

While most terminals can access LEO or GEO orbits, most can’t use both at the same time. “That has to change over time,” said Purohit. “Customers are increasingly looking for having both simultaneously.” And even if you have the capability, multi-orbit and multi-beam still have significant obstacles. “If you switch from LEO to GEO, will the cryptos lose sync, for example?” asked Purohit, highlighting concerns about potential security tradeoffs.

The two main use cases for software on the network edge are network management and onboard computing and processing, said Kasaboski. “We’re seeing a rise of interest in both,” he continued. “Depending on which you choose, the definition and even the location of the edge might be different.”

“Network management can often be done at the gateway on the ground,” said Kasaboski, and having updated edge processing capabilities would make satellite companies more attractive to customers. Better processing and management capabilities across the network generally make software-defined satellites difficult to compete with. “Within satcom, [onboard processing] is rising with the tide of software-defined satellites, but it’s always an arms race,” said Kasaboski.

SD-WAN could also be a networking solution to the complexity of multi-beam, as it manages the network in orbit, high-speed data rate transfer and adjusts capacity as needed. “SD-WAN, as I see it, is part of the simplification process for adding that behind all of those access networks,” said Tobias.

AI/ML at the Edge

All three experts agreed that going forward, next steps will include adding artificial intelligence and machine learning (AI/ML) to satellite and user terminal software. “Applying AI and ML at the edge, at the core of the network, essentially starts to take the human out of the loop and simplify,” said Tobias. “That operation is going to be the next step, as we move beyond this digital transformation.”

“We’re seeing it in all aspects of the satellite sector,” added Kasaboski, including “orchestration and manufacturing processes.” Use cases for AI will be vast, but finding the most relevant and useful applications will be key. Specific use cases will likely include having AI/ML monitor network systems and alert for needed maintenance before the situation becomes critical.

“You don’t want to overuse it,” said Purohit. “There are some problems you can solve quite effectively without bringing AI into the mix.” But the cases where AI will be applicable have the potential to make a world of difference in network management. “When it comes to any kind of management of the network, you might be able to get a better view of the overall picture of the network using AI and make adjustments which a human might take much longer to react to.”

For more on the advantages of SD-WAN, securing edge devices and the digital transformation of the satellite industry, listen to the full episode here.