The Quality of Service Challenge for NGSO Constellations
Until recently our industry thought that it could make money by connecting the unconnected. As a result, Non-Geostationary (NGSO) operators and startups have been struggling to find profits and diverted strategic plans from providing rural broadband to supporting government and military applications and connecting businesses and cruise ships. SES’ O3b and OneWeb are perfect examples of this.
Times have changed though. Recently E-Space announced $50 million in funding for its plans to launch some 100,000 satellites focusing on enterprise and government. Rivada Networks also announced to enter the space industry with 600 Low-Earth Orbit (LEO) satellites, targeting the same markets. Our industry is finally waking up to the reality that running a successful datacom business in space is about providing a unique service that complements terrestrial infrastructure, rather than a commodity service that’s guaranteed to ultimately compete with it.
Now that multiple companies are designing constellations targeted at government and enterprise markets, how can we differentiate between them? Comparing strategic plans, leadership teams, technology partners, and investors still works, but the quality of service is critically important for these business markets.
When satellites operate in NGSO, any connection to such satellite is lost every couple of minutes, up to an hour. NGSO satellites orbiting Earth are only visible for so long, which is why sometimes thousands of satellites are being launched to ensure that at least one satellite is always in view. When a connection is about to be broken due to a satellite disappearing beyond the horizon, the service is reconnected to another satellite until this satellite is also fading from view, upon which the process is repeated.
For a single node to stay connected to an LEO constellation operating at 1000 km altitude, this will result in over 200 connections being dropped and reconnected every 24 hours. For an entire constellation servicing many networks and nodes, we are talking hundreds of thousands of connections every single day. Add inter-satellite links and this number will only increase.
Looking at terrestrial services, there is no such challenge. For microwave or fixed-line telecom services, setting up the connection is part of the service provisioning. When the connection is lost during service operation, it is considered an anomaly triggering service tickets and potentially outage credits. Meanwhile, the service provider focuses on managing bandwidth for which multiple techniques and protocols are available. Some are time-based like time-division multiplexing (TDM), and while highly reliable, they have throughput limitations. Other techniques may lack the reliability of time-based systems, but by continuously adding bandwidth, reliability is highly improved and provides almost limitless throughput.
As a result, every terrestrial telecom service provider is happy to engage in detailed discussions with its (prospective) customers on the allocation and handling of bandwidth, on how quality of service is maintained and the underlying technologies that secure the link, the bandwidth, and the throughput.
NGSO operators will do all of the above, but in addition, there is this whole new frontier of how to manage thousands of connections that are dropped and built up every day. At the moment not much is being offered beyond saying that no connection will be dropped before a new one is setup and pulling data, i.e. “make before break.” Another one of my favorites is “TCP/transmission control protocol will take care of everything.” While the latter is usually offered with a wink, the former most definitely is not. When enough satellites are in view and when each of these satellites have sufficient spectrum, bandwidth, and processing power available, make before break can certainly be offered as a solution. Absent processing power and bandwidth, this approach will quickly turn from being a solution to becoming the problem.
In the context of quality of service, make before break also turns the service into a “best-effort service.” While there is nothing wrong with that – many terrestrial service providers take that very approach and continuously add bandwidth to address any issues before they arise – for space infrastructure it will put a very serious constraint on the number of customers you can take on and compels the satellite operator to apply considerable margins to accommodate for weather and potential traffic bursts.
Take SpaceX’s Starlink — with limited users, the system performs well and the data rates are impressive. To get to their goal of $35 billion in revenues however, they need to get to 30 million users and manage north of 5 billion(!) connections per day. The amount of bandwidth and processing power that will take is staggering and assuming it will all scale nicely is optimistic.
Unlike terrestrial networks, there is no easy solution to fix network hot spots. Any satellite flies over the empty Pacific Ocean as well as over New York city, turning the NGSO constellation into one single network without options to compartmentalize. In combination with the scarcity of frequencies, finite processing power and limited backbone capacity, throwing bandwidth at the problem now has limitations and will create a its own set of challenges. As a result, it seems that it is now on enterprise and government customers to ask detailed questions on how quality of service is handled, prior to accepting offers from NGSO operators.
There are also satellite examples of permanent connection bandwidth (satellites in GEO) as well as there are examples of make before break terrestrial services (cellular telephony). That said, neither of these services are meaningful in the context of multi-gigabit, low latency networking, relevant to business users.
With multiple aspiring NGSO operators now focusing on enterprise and government markets, we need to better understand their approach to meeting their customers’ quality of service standards. Merely relying on traditional terrestrial approaches comes with limitations and only solves part of the problem. Faced with the additional challenge of managing thousands, potentially millions of connections that need to be dropped and reconnected each day, these operators have an opportunity. Particularly for the enterprise market, the operator that has developed a robust and well documented system that guarantees quality of service by not only managing bandwidth but also these connections, is almost guaranteed huge commercial success relative to those who rely on the best-effort approach of make before break … or TCP for that matter!
Ronald van der Breggen is an independent consultant at Route 206, helping technology companies become commercially successful. His latest achievement was securing over $2 billion in customer commitments for LeoSat. Ronald holds a business degree from Nijenrode Business University and a master’s degree from Delft University of Technology, both in the Netherlands.
This article was first published on his LinkedIn.