Direct-Satellite Internet Makes Gains
Data networks in the sky are becoming more practical and cost effective, even against cable and DSL networks, with more breakthroughs just over the horizon.
Recently, the gap between direct-satellite Internet and cable/fiber optic in service quality in proportion to price has been closing. Most users should stay with cable for the next few years, but specialty users should take another look at satellite now, and plan to review periodically, as the gap continues to close.
In direct-satellite Internet, your router links to a satellite modem that transmits, via a dish antenna, directly to and from a communications satellite in geosynchronous orbit. That's the same orbit familiar from satellite TV; in fact, some satellites are multipurpose and carry Internet, television, and voice phone.
Direct-satellite Internet can be available in places where cables of any kind are physically impossible or economically unfeasible: ships at sea, remote military or scientific installations, or sparsely populated areas where construction is expensive or difficult.
In principle, direct satellite should have some major advantages. Much of the infrastructure can be installed right on the customer's property. The satellites relay via ground stations, so larger and more expensive pieces of infrastructure are safely on the provider's own company property -- no need to obtain rights of way or dig miles of ditches. The provider can serve anyone with a view of the southern sky. Best of all, the most expensive infrastructure is out of reach of weather, vandals, and uncooperative property owners.
For narrow niches such as offshore drilling rigs, cruise ships, little mountain towns, weather stations, geologists' camps, and so forth, those advantages are decisive. The cost: the key piece of infrastructure must be installed in a far away orbit, which can only be reached with millions of dollars worth of rocket. This has other negative implications: long latency (speed-of-light signal transmission delay) -- so fast, real-time interaction is unavailable; no service or maintenance by human crews, so that catastrophic failure would be far more catastrophic; and capacity issues that will remain important well into the 2020s if the providers stay with their present plans. A smaller nuisance is that connection is much more complicated than pushing a plug into a socket -- you need a technician to aim your antenna at the right part of the sky.
Many problems remain and some problems have no solutions -- for latency, you're up against Einstein. Yet great progress has been made in the last two years. Satellite broadband is now highly competitive, and growing more so in many remote, difficult areas where until recently, the alternatives were very slow DSL, legacy dial-up, or doing without an Internet connection.
Under a new generation of powerful satellites that use spot beams and the high-frequency Ka band, prices have fallen a long way. Monthly subscription fees now are generally under $60 for consumers. Download speeds are now 12 Mbps to 15 Mbps, and upload speeds are well above 1 Mbps.
Unfortunately, while they'll always sell you the fastest speed they've got, total data throughput is severely constrained, with various plans capping it at 10 GB, 15 GB, or 25 GB a month -- not a lot if you're receiving large amounts of video and audio, or if you need to do extensive, real-time interaction with a simulation or a large number of sensors. Unlimited downloads are available in a batch environment during off-hours from Exede, ViaSat's service, but it's doubtful that genuine, no-data-limits satellite broadband will happen until at least the next generation of satellites.
Still, the gap between fiber/cable and satellite is closing, and for some specialty users, the gap may have closed enough to make it worthwhile.
Here's a look at where we came from, what has changed recently, what changes are coming, and a couple of problems that will probably remain problems for quite a while.
In 1996, the first direct-satellite Internet service was introduced as DirecPC (now HughesNet). It was allocated only a few slots in the high end of the L and the low end of the Ku microwave bands, i.e. a cluster of frequencies around 12 GHz, which is far below the 20 GHz to 30 GHz of today's Ka-band satellite transmission. Low frequency physically implies three problems at once: wide beams, low bit rates, and big antennae.
All the same, it offered a then-awesome 400 kbps (compared to the 56 kbps of the dial-up services most people and businesses still had).
Things got better, but slowly. As recently as 2010, users who absolutely had to use direct satellite were stuck with a bandwidth limit of around 750 kbps upbound and 256 downbound.
All that changed overnight with the launch of ViaSat-1 in October 2011, which provided 140 Gbit/s service -- at that time more than all the satellites covering North America combined -- of Ka band (high frequency), organized into 56 narrow and spot beams. The following July, the launch of EchoStar XVII (by Hughes) added another 100 Gbit/s, organized into 60 steerable beams.
The electromagnetic spectrum is crowded and getting much more so. Direct-satellite Internet comes along after many generations of more and more technologies moving into the radio-through-microwave spectrum, and there just won't be very many new frequencies opening up anytime soon. Hence this workaround: steerable spot beams. Higher frequency signals like Ka band can be focused into tight beams with footprints a few hundred kilometers across, instead of the continent-sized areas covered by the older, lower-frequency beams.
This picture shows the results; different colors depict different frequencies, and if the spots are narrow enough, you don't need nearly as many frequencies to carry the same traffic. (You might recognize the solution to the four-color problem.) And because the beams are steerable, changes in usage, whether temporary or permanent, can be accommodated by moving beams from low to high usage areas.
Intelsat is aiming to enter the high-throughput broadband direct-satellite market by replacing its existing fleet with Intelsat Epic: C/Ku/Ka band satellites that have both wide and spot beams and 25-60 Gbps of throughput. The first two Epics, Intelsat 29e and 33e, are scheduled to go up in 2015 and 2016. Hughes's EchoStar XIX, with 160 Gbit/s throughput and 120 spot beams, is scheduled for launch in mid-2016. And Viasat-2, scheduled for 2016, is being kept tightly under wraps; reportedly, it will use a new technology to replace spot beams. All ViaSat will say is that they expect it will have 2.5 times the "equivalent" capacity of ViaSat-1, and its coverage area will include nearly all of North America and Europe.
But if everyone is telling the truth -- and they certainly all sound confident -- then satellite-direct broadband capacity should more than double in the next two years.
Worldwide satellite broadband is expected to quadruple before 2020, essentially using up all the new satellites and probably needing a few more. The advantage of high-speed Internet wherever you are, even if you're somewhere pretty hard to get to, is overwhelming. (Note, though, that the projected 6 million is less than 1% of the billion people now online.)
Direct-satellite Internet routinely delivers much more bandwidth than the provider promises. Typically they promise 12-Mbps downloads and deliver around 16; the promised 1 Mbps of upload speed is more often 1.5 Mbps or better. That's fast enough for nearly all business purposes, unless you're a gaming company, a defense agency operating drones, or a high-tech hospital doing surgery over telepresence.
As more traffic moves to higher frequencies, "rain fade" becomes more of a problem. Your microwave oven heats water efficiently because the water molecule is an electric dipole and high-frequency microwaves make it spin and absorb energy.
In exactly the same way, whenever your signal has to pass through rain or snow, a certain amount of the energy goes to warming up the precipitation. This loss can exceed 20% in a tropical rainstorm, or worse if your antenna is covered with snow. Solutions include increasing transmission power, various ways of switching from ground station to ground station to avoid storms, and putting your antenna where you can reach it with a broom.
Latency is the time between sending a packet request and the requested packet's coming back to the requesting server. It's commonly measured in milliseconds (msec); the lower the better. Some cable services offer latencies between 20 msec and 40 msec and nearly all less than 100 msec; most of that is processing time in the servers. But no matter what, direct-satellite Internet is stuck with a latency of about 500 milliseconds. Half a second is not a deal breaker if you're filing sales reports or even downloading video or audio, but it's a major nuisance if you're sending control directions for a drone or robot, trying to collaborate on a live musical piece, or playing a multi-player first-person shooter.
Why the huge difference? Because to maintain fixed positions in the sky, satellites have to be in geosynchronous orbit, about 35,800 kilometers above sea level at the equator. When you send a packet request, it goes out your antenna, 35,800 kilometers into space, comes about the same distance back to a ground station on Earth, then through a short pathway of broadband cable to the requested server. The packet then retraces the same path. That's between 140,000 and 150,000 kilometers. That's just under half the 300,000 kilometers that a signal can travel in one second at the speed of light (which, as Einstein taught us all, is not just a good idea -- it's the law).
By contrast, if you're mostly talking to servers on your own continent via cable, that's a round trip of a few thousand kilometers at worst -- and it only takes a third of a millisecond for a packet to traverse 1,000 kilometers.
The satellite latency issue is tied to the use of geosynchronous orbit, necessary for delivering an Internet connection from a single satellite. But suppose we weren't limited to a single satellite.
In spaceflight jargon, a constellation is a group of satellites working together. If you had enough satellites with their beam footprints overlapping on the ground, playing follow-the-leader in the same orbit, relaying signals back and forth to each other via enough ground stations, they wouldn't have to be way out in geosynchronous orbit. They could fly as low as practical, wherever there was an open orbit for them to fit into.
Currently under construction, the O3b constellation of 8 satellites will provide the rough equivalent of EchoStar XVII in capacity. But by working in tandem, they will be able to orbit at about 8,000 kilometers altitude, cutting that required Einsteinian transit time by more than two-thirds. That should bring direct-satellite latency down to the low end of what you would get with cable or DSL.
The satellite latency issue is tied to the use of geosynchronous orbit, necessary for delivering an Internet connection from a single satellite. But suppose we weren't limited to a single satellite.
In spaceflight jargon, a constellation is a group of satellites working together. If you had enough satellites with their beam footprints overlapping on the ground, playing follow-the-leader in the same orbit, relaying signals back and forth to each other via enough ground stations, they wouldn't have to be way out in geosynchronous orbit. They could fly as low as practical, wherever there was an open orbit for them to fit into.
Currently under construction, the O3b constellation of 8 satellites will provide the rough equivalent of EchoStar XVII in capacity. But by working in tandem, they will be able to orbit at about 8,000 kilometers altitude, cutting that required Einsteinian transit time by more than two-thirds. That should bring direct-satellite latency down to the low end of what you would get with cable or DSL.
-
About the Author(s)
You May Also Like