ALEXANDRIA, Va. — Data generated in space is growing rapidly. A typical commercial imaging constellation collects more than 100 terabytes of data per day. Space data traffic is expected to reach 566 exabytes over the next decade, according to NSR, with satcom accounting for over 90% of that total.
Edge computing on satellites and terminals is an emerging solution to manage the massive volume of space data traffic. (Source: Pixabay)
Edge computing, or distributed computing, has become a critical solution for managing and extracting value from the massive amounts of data generated terrestrially and in space. As satellite and cloud service providers strengthen their mutual relationship, edge solutions are emerging as a powerful enabler for satellite networks, from the terminal on the ground to the spacecraft in orbit.
The value of applying computation power at the edge of the network is evident to civil, commercial and military space leaders. U.S. Space Force CTIO Dr. Lisa Costa has remarked that “we need computational and storage power in space.” Meanwhile, industry leaders have begun to view cloud technologies, like edge computing and virtualization, as a necessary precondition to unlock the potential of multi-orbit constellations, proliferated LEO and high throughput satellites.
What Is the Edge?
At its most basic level, the edge refers to a point between two networks—such as a local network and the cloud or core network—or the point where a user or machine interacts with the network.
An edge device is any piece of hardware that enables physical connectivity and controls data flow between two networks. It can be a smartphone, a router connecting home devices to the internet, a satellite pre-processing raw imagery, or a terminal autonomously determining the optimal transport network for a satcom link.
Edge computing is all about placing workloads as close as possible to where data is being created or where the end user needs it. It seeks to address the problems of data in motion by limiting the need for long-distance transfers. Rather than backhauling large amounts of raw data to be processed in the cloud or data center, computing occurs on-site. This reduces strains on centralized networks, which can lead to delays and service disruptions. Depending on the use case, edge computing can involve data storage for later analysis, purging irrelevant data and running applications to synthesize insights and inform real-time decisions by machines or operators.
Edge computing is a fast-growing enterprise technology. STL Partners estimates the global edge computing market will reach $445 billion across verticals by 2030. Within the broader but related cloud computing market, satellite will represent a $16 billion revenue opportunity through 2029, according to NSR. Satcom will account for the lion’s share of generated revenue.
The Space Edge
In many ways, a satellite is the quintessential edge device. It is remote, often disconnected from a data center and generates loads of data. Outfitting it to function as an edge device, with local compute, unlocks a host of operational benefits, including reduced latency, greater security and lower transport costs.
The idea behind space edge-hosted payloads is to reduce the need for data to make multiple round trips to an Earth station or terrestrial-based cloud. It’s a long way from LEO, MEO or GEO back to Earth and even longer for analysts on the ground to crunch raw satellite data and route it back to the satellite.
Deep space exploration and permanent orbital outposts will rely on edge computing. Image shows Axiom Space’s planned commercial space station. (Source: Axiom Space)
“There are significant benefits to moving the edge into space, including latency, security and cost,” explained Dennis Gatens, CEO and Founder of LEOcloud, a space-based Infrastructure-as-a-Service (IaaS) company that provides multi-cloud edge computing services.
“Reducing latency is of great benefit to the end user,” he told Constellations. “Instead of incurring the delay and cost of moving high volumes of raw data back to Earth, the processing can be moved into space, including to the point where actionable insight can be delivered directly to the end user without the raw data touching Earth.”
Edge computing reduces the time between data collection and decision-making, which is critical to latency-sensitive applications, like communication and autonomous systems. This feature is critical in the growing field of onboard AI/ML applications, such as navigation, analysis of satellite health and space situational awareness. In EO and remote sensing satellites, onboard processing allows operators to downlink refined information, saving time and bandwidth.
“There's a rapidly emerging need for commercial, government and military customers to have this capability as close as possible to the sources and users of data,” Gatens said.
This need will continue evolving as humans move into cislunar and deep space, where systems must become less dependent on regular support from Earth. It will also be a core capability for future permanent orbital outposts, such as the commercial space station under development by Axiom Space. LEOcloud and Axiom recently announced an agreement for cloud and edge services to support the commercial space station, which is scheduled to launch its first module in 2025.
The Edge of the Satellite Ground Network
In the ground segment, the edge of the satellite network refers to the terminal, or in some cases, the teleport or gateway. An edge can be on a plane, boat, vehicle or other remote site. It’s often at the farthest point from the core of the network, closest to where data is generated.
Just like the space edge, edge computing at the terminal improves network performance by moving workloads closer to the end user, reducing latency, improving security and mitigating the cost and strains of moving data to a centralized network.
Beyond processing workloads, edge computing can be used to host various applications, while allowing users to exploit the benefits of software-defined networks (SDN) and network function virtualization infrastructure (NFVi). These agile networks and virtual functions are gradually replacing traditional satellite network architectures and hardware components.
An edge device with general compute can run virtualized network functions (VNF), like virtual modems, virtual firewalls, virtual network accelerators and SD-WAN/SASE. These software functions introduce capabilities at the edge that otherwise would require power-intensive, dedicated hardware to support.
“By leveraging edge compute and network function virtualization at the edge, we’re able to dynamically adapt in real-time as conditions in the network change or as business needs evolve,” explained Kevin Tobias, Director of Product Management and Edge Products at Kratos.
Edge processing can enable operators to make on-the-fly changes to the mission at a remote terminal. Image shows the ALL.SPACE satcom terminal. (Source: ALL.SPACE)
An example of this is resilient communications in a contested, congested or degraded RF environment. Using general purpose edge compute and NFVi, a user can spin up virtual modems or change waveforms on the fly to avoid interference and protect critical communications. This can enable users to switch between multiple orbits or networks and utilize the best-performing link—whether it’s LEO, MEO, GEO or a terrestrial cellular network.
“A user can basically create rules at the edge to orchestrate new applications and provide new services without having to actually touch the site,” Tobias added.
Similar to the space edge, there are numerous use cases for ground-based satellite edge processing. They can range from satcom and MILSATCOM to Earth observation, IoT and other value-added space services, including ML/AI-enabled applications.
The British terminal manufacturer ALL.SPACE incorporated edge computing on its multi-orbit, multi-network, multi-link satcom terminal. The edge device detects network interference and intelligently switches to a different link. Other terminals boast edge compute capabilities for automating and improving operator decision-making, managing network traffic and aggregating and backhauling sensor data within the Internet of Things.
What’s the Role of the Cloud?
Though they often appear together, the cloud is not the edge and the edge is not the cloud. Computing happens in both locations, but like real estate, the key differentiator is location—i.e. a centralized or hyperscale data center versus a point close to the end user or data generation.
The cloud facilitates two-way communication with the edge and establishes an environment where collaboration can occur more easily across locations, applications and network functions. Data backhauled to the cloud can be processed, stored or accessed by approved users within that environment. It can deliver commands, such as software updates or configurations, back to the edge. As the technology evolves, more enterprises are gravitating to edge cloud (a smaller version of the centralized cloud) and edge servers (essentially micro-data centers). These offer faster analysis and more powerful processing without connecting to a traditional cloud data center.
The wealth of data from space has created both the demand and incentives for hyperscale cloud providers to extend their services and service environments beyond Earth. LEOcloud is helping create seamless cloud environments from Earth to space. The space-hardened bare metal platform has scalable compute and data storage resources for multi-cloud hosting, enabling a hybrid cloud region from a satellite to a terrestrial data center—or future space data center.
“From an end user’s perspective, we’re just like the cloud infrastructure on Earth,” Gatens explained. “We look like a routed endpoint on a network just as you would on Earth and that’s the direction that the communications infrastructure is going in space.”
Digital ground service providers are similarly extending virtual cloud environments so that satellite services (satcom in particular) become just another node in a broader communications network. Workflows and functions previously siloed within a satellite operating environment are being virtualized and migrated into cloud and edge environments.
In a recent Constellations podcast, Amdocs Product Marketing Director Luc-Yves Pagal Vinette explained how the goal of cloud-native functions is to “break up the boundaries between verticals,” like those between the satellite and telecommunications industries.
Explore More:
Podcast: NFV, SDN and Enabling Dynamic Satellite Ground Networks
Satellite Industry Leaders Discuss Cloud Adoption, Evolution of Edge Computing
‘Cloud Is Eating Across the Entire Satellite Value Chain’: AWS and NSR Weigh In
Podcast: Automation, Service Orchestration and the Future of the Teleport
