To most observers, the idea of using satellites to cover wide areas of sparsely inhabited areas has an immediate appeal when compared to digging trenches or erecting poles to reach every underdeveloped nook and cranny in our vast country. It’s even more attractive in countries where there’s no wired infrastructure to speak of.
The new satellite networks being proposed
(Elon Musk’s Starlink being only the first of the contenders), all are taking the same fundamental approach, so network designers must understand the new technical variables they will be dealing with.
Let's review the basics of satellite communications. Communications satellite systems come in two main varieties: geosynchronous (GEOS) and low earth orbit (LEOS). GEOS systems face an insurmountable limitation with latency or delay. Hence virtually all of these new satellite options are based on the LEOS technology. Some one-off satellite solutions look to incorporate 5G cellular technology, but we will save those for the next installment.
LEOS-based systems use a very complex radio network in combination with packet switching technologies that require buffering that can introduce variables in terms of data rates, latency, packet loss, and potential limitations in terms of service availability, consistency, and reliability. In short, everything that will impact user experience (UX) for your work-from-home population.
What Happened to Geosynchronous Satellites (GEOS)?
I started working with satellite technologies early on in the mid-1970s, when all communications satellites (except Echo
) were of the geosynchronous variety (GEOS). GEOS describes satellite systems where the satellite orbits in a precise slot over the equator at an altitude of about 22,300 miles; at that altitude the satellite will orbit the earth every 24-hours. In this arrangement, a fixed earth station can be pointed at one point in the sky, and the rotation of the earth will keep the earth station pointed at the satellite continuously—hence the name “geosynchronous.”
While the orbital physics is inspiring, the GEOS service is not. Sticking that satellite 22,300 miles away meant any signal passing over the link would have to travel at least 44,000 miles, and has a major impact on latency. Even with radio signals that travel at approximately the speed of light, it will take a signal about a quarter of a second (i.e., 250 milliseconds) to traverse the link, so that's a round-trip delay of half a second or 500 milliseconds.
That 500 millisecond-delay makes satellites virtually unusable for real-time business voice or video service. To put the problem in context: for business voice calls, we are looking for one-way latency under 150 milliseconds, and less than 100 milliseconds would be even better. We should note that this delay is also detrimental to data applications, and you can forget about any very-low latency requirements like online gaming.
Even in the 1970s, customers quickly realized how important this latency issue was to voice and data service quality and began insisting on terrestrial rather than satellite routing. GEOS were then relegated to tasks that were not as latency dependent, like one-way TV distribution and VSAT
point of sale systems for retailers.
Latency Problem Leads to LEOS
Based on questions I get from clients—many people still appear to be stuck on the idea that all satellites are GEOS and immediately dismiss any satellite solution due to the latency consideration. However, the LEOS technology is another beast entirely.
The engineers figured out early on that we weren’t going to accelerate the speed of light, so the only way to reduce the latency over a satellite link would be to fly the satellites closer to the earth. That idea was first commercialized by Motorola and used to deliver Iridium
, the first Low Earth Orbit Satellite (LEOS) system that went commercial for voice service in 1998.
To cover the entire planet, the Iridium network consisted of 66 satellites orbiting at an altitude around 485 miles. The lower altitude reduced the latency, and importantly, the shorter transmission range required less transmit power and hence improved battery life, a key factor as Iridium was designed to support mobile users. The Iridium satellites are in polar orbits and arranged in six orbital bands, with 11 satellites in each band. An Iridium satellite will be in view of a stationary ground station for roughly seven minutes.
The magic is that as each satellite moves out of range, the connection is handed off to the next satellite coming into view. The whole idea is much like a cellular network, only here, the users essentially stand still, and the cell sites pass by overhead.
The New Variable: Satellite Cross-Links
While that is the basic picture, it is important to recognize that there are two distinct networks involved in delivering a LEOS-based service. One describes downlinks or how the satellites talk to the subscribers and ground stations, and the second describes the cross-links or how the satellites relay traffic to one another.
In Iridium, there are four designated downlinks worldwide for connecting the service to the wired telephone network; most calls are between Iridium subscribers and PSTN or cellular subscribers. When an Iridium user dials a connection to a user on the wired (or terrestrial cellular) network, whichever satellite passing over the caller receives the connection request, and then establishes a satellite-to-satellite connection to the satellite that is passing over the designated downlink station. A gateway in that ground station, in turn, establishes the connection through the wired telephone network (i.e., the PSTN) to the requested wireline or cellular subscriber. Calls to Iridium subscribers are beamed directly down to the subscriber’s handset.
If any Iridium call lasts longer than the maximum seven minutes that the satellite might be orbiting over that Iridium user, the caller’s connection is handed-off to the next satellite coming into view, and the satellite-to-satellite connection to the downlink satellite is rebuilt.
Iridium's major downfall is it was designed for real-time voice with constant delay. The company later introduced a data service capable of downstream rates under 1 Mbps using its second-generation satellites. However, that won’t meet anyone’s expectations for real wireless broadband.
Starlink’s roots go back to 2014 when Elon Musk‘s SpaceX filed a stealth application with the ITU under the name STEAM. SpaceX eventually came clean on the STEAM relationship and launched its first Starlink satellites in 2018. The long-term plan is to have three orbital “shells” with 1,440 mass-produced satellites at 340 miles, 2,825 at 690 miles, and 7500 at 210 miles; the initial deployment is at the 340-mile altitude. All connections will be end-to-end encrypted.
While the idea of a LEOS constellation was pioneered by Iridium, Starlink took the idea in a much different direction. First off, where Iridium was designed primarily for real-time voice calls, Starlink is a data network first and foremost. It is reportedly using a simplified version of IPv6 for routing. That also means that the Starlink satellites will need the ability to buffer traffic, which can introduce variations in delay and potentially dropped packets. That buffering will be required as the satellites routinely rebuild connections every few minutes. Starlink is also putting other new ideas into orbit.
The Starlink network will use 12 GHz radio channels for uplink/downlink connections, but the inter-satellite links are planned to use free space optics (i.e., fiber optics without the fiber). That idea had been proposed for iridium, but Starlink is making it happen.
What Starlink Delivers
Trial users for Starlink’s residential service buy and install a $599 terminal that includes a pizza box-sized dish that must be installed outdoors; there is a higher-capacity business model that costs $2500. The monthly charge is $110 for the residential service and $500 for the business offering. While it is primarily a fixed location service, Starlink does have a maritime capability for $5,000 per month that requires $10,000 in hardware. There are also fledgling plans to provide service to aircraft in flight. Starlink service has no data caps and no contracts.
Starlink promises to deliver downstream data rates of 50 M to 200 Mbps for residential users and 100 M to 350 Mbps downstream on its business offering. The company promises 20 M to 40 Mbps upstream; latencies in the 20 to 40 msec range are also promised. The maritime service promises potential data rates of 350 Mbps downstream.
In current testing by Ookla
, Starlink blows the other satellite offerings away both in terms of bit rates and latency, though it is still coming in at the low end of its own promised performance. Trial users see median downstream rates over 90 Mbps in the U.S., but that is still below the 100 Mbps downstream rate the FCC considers true broadband. The latency beats Starlink’s own projections and obliterates the other options running over geosynchronous satellites.
In short, if you’re looking for a satellite-based broadband Internet WFH option today, Starlink is the only one to look at. We will be tracking developments and reporting on the other planned satellite constellations as they go into service to see how their performance stacks up.
The FCC’s Surprise
While all of the proposed networks employ LEOS technology, the initial market interest has focused on the Starlink because it is the only service actually up and running—if only on a trial basis. However, the FCC recently dealt the company and the technology a significant financial hit and a major vote of no-confidence.
Ms. Rosenworcel (clearly) has a different view of Starlink’s longer-term prospects of delivering on their promises than I do. However, her comment regarding the $600 cost of the Starlink terminal smacks of some kind of class warfare . In any event, pulling the plug on this technology before the initial deployment is complete should bring into question the entire idea of government paying carriers to spur deployments.
Meantime, it appears I’m not alone in questioning Ms. Rosenworcel’s judgment on Starlink’s potential. Soon after the announcement, FCC Commissioner Brandon Carr responded
to the decision to rescind funding, calling it a “clear error.”
I had offered conditional hope that RDOF could be a positive step when I wrote about it in 2020
, though I did reference “the government’s putting a benevolent thumb on the scale to help get the ball rolling.” I guess that benevolence was short-lived, and that thumb has now moved to the throat. Take this as a cautionary tale as to why the U.S. broadband business has done so well with less, not more, government involvement.
Other Starlink Challenges
Starlink (and, in the longer run, the other proposed satellite constellations) faces other potential challenges besides the ruckus around the 12 GHz band. Space junk, or the amount of retired space vehicles that continue to speed around the earth in various orbits, is raised frequently. The company published a technical paper on its plans to avoid collisions
; and claims that failed and retired satellites will deorbit within five years without propulsion and burn up on reentry.
There is also the issue with the impact on astronomy and other ongoing missions in space like the Hubble Space Telescope (which orbits at almost the same altitude as the current Starlink deployment) and the International Space Station.
How Does Starlink Service Stack Up for WFH?
Neither Starlink nor the cellular carriers’ FWA offerings come close to the near gigabit performance of the fiber or cable alternatives, so in the big picture, a wired connection is still your best option in terms of performance and availability—if you can get it.
As I noted in my piece on cellular FWA
, the cellular carriers are delivering downstream data rates in the 100 M to 200 Mbps range (compared to Starlink’s current 90 Mbps rate). However, the tests on the cellular network do not specifically break out indoor performance. That means we’re still guessing what data rate real-world cellular FWA users might get. By incorporating an outdoor antenna as part of the configuration, Starlink has tackled the problem of getting a good receive signal, the single biggest performance factor in any radio system.
While Starlink has a clear lead in the satellite space, what we haven’t seen from the company (yet) is a track record of consistent reliable performance of the service it has promised in the face of weather conditions, routine failures, and possible quirks of geography that may limit availability in valleys or other challenging settings. Right now, the network is still performing at the low end of its promised data rates, with only 400,000 trial users. On the other hand, Starlink has not yet gotten to it planned full deployment, so its long term capabilities are still to be determined.
From a business standpoint, Iridium and its competitor Globalstar have struggled to make a profit at providing wireless voice services to sparsely populated areas. The problem with providing service to the middle of the desert is that there’s usually no one out there! Possibly the strategy of covering every remote area (including the oceans) with a voice-capable broadband data service might be the key to turning that prospect around. In 2017, Musk projected revenues of $30 billion from Starlink by 2025 versus only $5 billion from its space launch business .
The other big factor driving the satellite constellations will be competition. Even if the FCC is right and Starlink can’t deliver on its promised data rates, maybe one of its competitors will. This new generation of data-oriented LEOS networks involves putting a lot of new technology (like IP routing and free space optics) up in space, so we shouldn’t expect the engineers to get it 100% right on the first shot.
As if there haven’t been enough surprises in this story. As I was wrapping this up, T-Mobile made a joint announcement with SpaceX
, describing their plans to have the Starlink satellites communicate directly with cell phones to provide mobile coverage over the entire planet for T-Mobile’s cellular subscribers. This initial plan is to deliver just text messages and eventually move on to the full range of voice/data/video services.
Like many “new” ideas in wireless, this one has been making the rounds for decades (and there are at least three other start-ups currently working on the same idea). The plan predictably stumbles each time on the same fundamental capability: “How do you make a cell phone designed to send a signal a couple of miles to a cell tower miraculously send a signal several hundred miles into space?”
Without a solution to basic challenges like that (and that’s just the opener)—this is little more than an overly optimistic press release. In short, a full analysis of this announcement must wait for another post. Keep an eye on SpaceX in the meantime. But don’t hold your breath waiting for them to deliver on the mobile cellular part of the plan.
This post is written on behalf of BCStrategies, an industry resource for enterprises, vendors, system integrators, and anyone interested in the growing business communications arena. A supplier of objective information on business communications, BCStrategies is supported by an alliance of leading communication industry advisors, analysts, and consultants who have worked in the various segments of the dynamic business communications market.