Why do Starlink satellites orbit so fast around the Earth, approximately every 90 minutes or so? Why are they not in geostationary orbits?
$\begingroup$
$\endgroup$
Add a comment
|
6 Answers
$\begingroup$
$\endgroup$
1
From the Wikipedia article on Geostationary orbit:
A geostationary orbit, also referred to as a geosynchronous equatorial orbit (GEO), is a circular geosynchronous orbit 35,786 km (22,236 mi) in altitude above Earth's equator, 42,164 km (26,199 mi) in radius from Earth's center, and following the direction of Earth's rotation.
That one sentence says all the good and bad about GEO vs. Low Earth Orbit (LEO).
- 35,786 km (22,236 mi)
That is a long distance. That means on the equator, directly underneath a satellite, it will take almost 1/4 of a second just to send a signal from your device to the satellite and back down to a nearby base station. If you are not near the equator it is a longer distance. If you are on the equator but not directly under a satellite then it is a longer distance.
Compare that to satellites in LEO. According to this space.com article, a typical orbit is 550 kilometers (342) miles. That's less than 2% of GEO. In addition, while the satellites are constantly moving around the Earth, there are around 6,000 of them (and more launched almost every day!) so there is one, relatively speaking, close to overhead at any time.
- Geosynchronous
This is the good part of GEO. The satellite effectively stays above one spot on Earth. That is great because it means your satellite dish doesn't have to move to track the satellite. But that doesn't help so much if your satellite dish is on a moving car, truck, ship or plane.
So why aren't all communications satellites LEO? There are two key changes in the past few decades that have made LEO more practical for a constellation of communications satellites:
- Improvements in Computers
Yes, really. The first satellites literally reflected signals. They quickly progressed to amplifying signals and eventually to actually processing and directing the resulting data to specific locations. But it has taken decades of improvements to get to the point where an extremely sophisticated computer (and related radio transmitting and receiving hardware) capable of handling thousands of simultaneous high-speed connections could be launched in a relatively small satellite. That is critical because the target market is not occasional or emergency telephone use or low-bit rate (by low, I mean kilobytes per second - think 1990s wired modems at best) as was the case with earlier systems such as Iridium, but high-speed connections comparable to cable or fiber internet. Other satellite companies have progressed similarly, but this is a change that favors LEO as lots of fast connections works well with the low latency of LEO to provide lots of fast usable connections.
But in addition to each satellite being far more powerful than previous generations, you also need a lot of satellites, which means launching a lot of satellites. Three geosynchronous satellites, equally spaced, is enough for global coverage. With LEO you need more satellites. The Iridium satellites were named after 77, the atomic number of the element iridium and the original planned number of satellites. But the more satellites you use, the smaller the area each satellite communicates with and the higher total available bandwidth. Starlink is currently up to about 6,000 satellites, and the eventual total is planned to be at least 12,000 satellites, possibly a lot more. Satellites get cheaper with mass production, just like TVs and computers and other electronics. But rockets...
- Low-cost Launches
SpaceX has pioneered the reusable rocket. That, combined with other improvements, has dramatically lowered the cost per satellite launch compared to previous options. In addition, SpaceX is vertically integrated - they are building Starlink satellites and launching them on their own rockets, so even though the launch costs have to be accounted for internally, they don't need to make a profit on Starlink launches. The result is that while their cost per satellite launched is not public, it is certainly far below the competition.
answered Jun 10 at 16:33
-
8$\begingroup$ Also, a geosynchronous satellite covers a wide area of the globe, depending on the design of aerial. So you only get a very small amount of bandwidth, ok for legacy TV or voice calls from a few ships, but not for 300Mbits to millions of people. $\endgroup$– RichCommented Jun 10 at 21:30
$\begingroup$
$\endgroup$
4
The low earth orbits mean that they are more responsive. The round-trip from earth-to-GEO-to-Earth is typically around a half a second. The LEO delay for Starlink is closer to 1/10th of a second. That means that the user doesn't feel the lag. Their interaction with a website is closer to what the typical high speed internet user feels.
Remember a big part of their market is for users who don't have access to high speed internet from a terrestrial system. The GEO based solutions in the past had too much lag according to users.
Once they decided to use LEO then number of satellites went up. In fact it went up a lot. The advantage that they had was access to a reusable launch system that they could influence the design specifications of the rocket.
-
$\begingroup$ Regarding the latency - it turns out that QuickAssist (remote access software) has trouble doing a handshake/connecting over Geo-based satellite internet, as best as I can tell, due to the latency. $\endgroup$ Commented Jun 10 at 18:48
-
7$\begingroup$ The roundtrip to Starlinks in LEO is actually even less than that. The actual signal takes ~2ms to reach a satellite if it's directly overhead, but with signal processing included the typical latency to your destination server is somewhere between 20 and 50 ms. That's Earth>LEO>Earth>Fibre backhaul>Server. If there's in-space sat-to-sat hops it can be a bit more. All that to say that for LEO, light speed is not limiting. For GEO it is. They're working on improving the routing speed. $\endgroup$– MqriusCommented Jun 10 at 20:07
-
12$\begingroup$ Besides shorter latency, there is a tremendous advantage in power requirements because of the shorter path length. Remember the inverse square law! $\endgroup$ Commented Jun 10 at 20:16
-
1$\begingroup$ @PhilFreedenberg Interestingly, the sum of the many LEO-satellites' power should resemble the one GEO satellite's power if the per-antenna-on-the-ground requirement is the same! (You know, shells of spheres ... ;-) .) $\endgroup$ Commented Jun 11 at 11:21
$\begingroup$
$\endgroup$
4
There are advantages and disadvantages to placing satellites in a LEO orbit. Let's start with the advantages...
Advantages
The advantages of a LEO constellation over a Geosynchronous satellite are:
- A satellite in Low Earth Orbit (LEO) can spatially multiplex better.
What this means is that the system can divide the territory below it up into relatively small cells. Then it can communicate with individual cells simultaneously, by using its directional antennas to place multiple beams on the ground. I believe that this image from starlink.com attempts to illustrate this concept. Each white dot on the US represents a beam from a satellite traveling across the sky somewhere above it. One satellite can place several beams simultaneously. (It's tricky to understand what the image represents, so I circled one satellite in orange and showed the four beams within its cone using red lines.)
Beams are good for both transmitting and receiving. We tend to think of beams as being used for only sending light or information, but, in telecommunications, the idea of "a beam" can also refer to listening for a signal coming from a certain specific direction. So, a white dot represents both a place where the satellite is sending its data and a specific place where it is looking for return signals.
A geosynchronous satellite can do the same trick, but because it is much further away, its "white dots" will be larger, so it can't maintain nearly as many simultaneous conversations with transceivers on the ground as a LEO constellation of satellites can.
More simultaneous beams placed in a given territory means more parallel communications links which in turn means more total bandwidth. More bandwidth translates to more customers and more revenue.
- It's less costly to place a satellite in LEO than in GEO.
Payload to LEO for a reusable Falcon 9 averages around 9500 kg but it has achieved a record of 17500 kg. However, it can only place 5500 kg on a GTO orbit (ref). Then additional delta-v is required to circularize the GTO orbit into a GEO orbit. This means that some of that payload will be consumed by a thruster and its propellant.
- Latency is improved.
Overall latency is made up of many components including time spent traveling through the user's equipment, time-of-flight to the satellite, through the satellite, back down to the ground station, and then time to the point of presence (where the ground station connects to the internet), then through the internet to the server that the customer wants to interact with, and then back through the same path to the customer. Placing the satellite closer to the customer reduces the time of flight between earth and the satellite. This amounts to a time saving of roughly...
$$t=2*35000000/c = 2*35000000/299792458 = 0.23 seconds$$
While this is an improvement, it is not the real reason why Starlink can provide better service to more customers than satellite internet providers that use geosynchronous satellites. The real reason has more to do with the fact that the customers of Starlink's competitors are all competing for a small share of a limited amount of total bandwidth, which has more to do with reason 1, "spatial multiplexing", above.
Disadvantages
There are several disadvantages to LEO satellite constellations too, including the extra expense of user equipment that is capable of tracking the satellites, the need to deal with obstructions that prevent the user equipment from having a clear view of the sky, the need for the satellites to cycle power through their batteries 5840 times per year, satellites being idle whenever they are not over territory with paying customers, the higher overhead cost of placing enough satellites in orbit to provide customers with continuous service, and the service quality's sensitivity to single satellite failures. But overall, the constellation's advantages (mainly #1) outweigh its disadvantages.
Conclusion
So satellites orbit faster when they are in a LEO orbit, which has a much shorter orbital period than a GEO orbit. Starlink satellites are placed in a LEO orbit mainly because closer proximity to Earth allows more communication to happen in parallel, which leads to greater total bandwidth, which leads to more customer revenue.
-
3$\begingroup$ A GEO satellite would also need to be bigger and heavier since it would need a stronger transmitter/receiver and hence more power. So you wouldn't be able to launch as many per rocket. GEO is also vulnerable to shadowing - if there's terrain or a building between you and the satellite, you likely can't communicate with it. This gets worse the further from the equator you go since the satellite gets lower in the sky. It can be mitigated by using more satellites in inclined geosynchronus orbits or Molniya orbits, at greater cost. $\endgroup$– ÅsmundCommented Jun 11 at 8:05
-
-
$\begingroup$ I am a bit surprised by your statement that the "the service quality's sensitivity to single satellite failures" is a disadvantage of LEO constellations. Do you really think that, say, Starlink is more sensitive to single satellite failures than, say, Inmarsat? $\endgroup$ Commented Jun 17 at 21:40
-
$\begingroup$ Yes, I can see how that statement would be confusing. Satellite internet service needs to be reliable to meet customer expectations cultivated by terrestrial technologies. Without redundant coverage, if Starlink loses one satellite in its constellation, every customer will experience periodic loss of service until the hole in coverage is plugged. But, the same is true for GEO sats. The probability of sub-par service is 1-(p^n) where p is the probability that any single sat is functioning and n is the number of satellites. As 'n' is large for LEO constellations, 'p' must also be closer to 1. $\endgroup$– phil1008Commented Jun 18 at 6:33
$\begingroup$
$\endgroup$
1
Another advantage of LEO compared to GEO, related to distance, is signal strength.
Unless you use laser (which would mean very very accurate pointing, both ways), when you send a signal, the strength of the received signal decreases with the square of the distance.
This means that a signal received from 40 000 km away is going to be thousands of times weaker than the same signal received from 500 km away.
This in turn means one or several of:
- Higher possible bandwidths
- Smaller antennas
- Lower power
- Reduced requirement to point precisely at satellites and/or track them
- If you're lucky, ability to use without line of sight (indoors / in urban locations)
The exact combination is a trade-off based on the priorities of the operator.
There's an analogy with Wi-Fi vs. cellular: at shorter distances, you can achieve higher throughputs than at larger ones.
-
$\begingroup$ Laser also loses signal strength with square of distance. You just start with a much lower dispersion, so the factor is much lower, but it's still the same surface-by-distance law. $\endgroup$ Commented Jun 13 at 22:09
$\begingroup$
$\endgroup$
10
If they were in geostationary orbits, signals would take just over a tenth of a second to get all the way up to the satellites, and then have to be transmitted all the way back down again. In addition to the time it takes to send the signal up and down, the farther out you have your shell, the longer the path the signal will have to take; since the shortest path will be a great circle.
iirc, at their current orbit height, starlink barely beats out longer undersea cables in terms of speed, as the speed of light through a vacuum is faster than through optic fiber.
-
1$\begingroup$ This is not the main reason. Starlink is worldwide broadband service, if they had them in Geosynchronous orbit, they can only service one area. (Most communication satellites are actually not in Geosynchronous orbit) $\endgroup$ Commented Jun 10 at 13:11
-
$\begingroup$ @LawnHollanderLawn they have a large fleet. Are you implying they would have difficulty serving polar regions? $\endgroup$– BasilevsCommented Jun 10 at 13:21
-
3$\begingroup$ @guest4308, I find the clause about undersea cables highly suspicious. There is no way for a radio or laser channel to be anywhere near efficient as a dedicated fiber. Also, Stalink uses those cables for ground stations. $\endgroup$– BasilevsCommented Jun 10 at 13:23
-
-
3$\begingroup$ @Basilevs OK, but "...in terms of speed, as the speed of light through a vacuum is faster than through optic fiber" is clearly about "the speed of light through a vacuum" vs "optic fiber" which goes to latency, not bandwidth. The answer has only two sentences, and the first sentence introduces latency: "... the time it takes to send the signal up and down..." as well. Latest designs of long-haul fiber & fancy modulation schemes push >Tb/sec over long distances, but Starlink doesn't serve whole cities. It fills the gaps where folks are far from those major trunk feeds; with competitive latency. $\endgroup$– uhohCommented Jun 10 at 22:17
$\begingroup$
$\endgroup$
Long ago I worked with satellite internet because it was the only thing available. After a whole lot of digging I finally figured out the reason e-mail was utterly broken was that the protocol required acknowledging every line. Ping was IIRC in the realm of 800ms. There was a hardcoded 5 minute timeout for downloading a message. That's 300 seconds, or 375 round trips if my memory of the ping is accurate. Any message beyond 375 lines could not be downloaded--and the offending messages had images encoded within.
Anything working through geosynchronous satellites will have the same insane ping.
