100/16 Mbps is pretty decent I guess, hopefully it doesn't go down as the number of users goes up. The latency is great imo, 40ms using satellites? I don't think anyone has achieved that before.
Would a bigger dish work better or not?
100/16 Mbps is pretty decent I guess, hopefully it doesn't go down as the number of users goes up. The latency is great imo, 40ms using satellites? I don't think anyone has achieved that before.
Would a bigger dish work better or not?
A bigger dish would not lower latency but may increase signal strength leading to better throughput. But it's not a simple parabolic antenna. It is the first consumer oriented phased array antenna tracking the satellites as they move across the sky, so that would increase the antenna cost even more.
A better signal strength would probably lead to better modulations being used, therefore, less transmission and reception time, and a better latency, or am I wrong ? (at least that's the case for Wi-Fi: the better the signal strength, the lower the transmission time so the better the latency).
At 100 Mbps with a 40 ms latency, there are about 2 megabits in the air between the ground station and CPE.
Is it though? Starlink orbits at 550 km, time-of-flight from ground to satellite to ground would be only 3.7ms, twice that makes ~20% of the roundtrip latency.
In this phase Starlink uses 72 orbital planes, with 22 satellites per plane, so 1440 satellites in total (they're almost there). It orbits at 550km above Earth's surface, so the orbit has radius 6921km, which gives an orbital length of 43486km.
Separation between orbital planes varies depending on your latitude, but assume the worst case, where it is 43486km / 72 / 2 = 302km¹. Thus, the nearest orbital plane is at most 302km / 2 = 151km away from the orbital plane directly overhead. However, since the planes process, on average the nearest orbital plane is only half that, or 76km away from the plane overhead.
Satellites within each plane have a separation of 43486km / 22 = 1976km. Thus, there's always a satellite at most 1976km / 2 = 988km away¹ from any point in each orbital plane, and on average there's a satellite half that away, or 494km.
Adding all this together, the nearest satellite is on average √(550^2 + 76^2 + 494^2) = 743 km away (at the worst latitude).
[EDIT: Actually, that's improper averaging, the correct average is obtained with ∫√(550^2 + x^2 + y^2) dx dy / ∫ dx dy on x=0..151, y=0..988, which yields 777km].
The original plan used 24 planes with 66 satellites, which reduces average distance to 617km. At more favorable latitudes the difference with the current design would be even larger.
[EDIT: This should be 635km.]
¹ This is distance on the surface of the orbital sphere, straight-line distance is a bit less. It probably doesn't make much difference.
From what I read it has to do with the fact the antennas are high gain directional antennas and not omnidirectional ones like on cell phones. With cell phones you are kinda walking around in a soup of cell signals all sharing the same spectrum at once... you and hundreds of other people are broadcasting in the same frequencies at the same time and they all tell each other apart because they all use a different “language”; the Wikipedia CDMA article does an excellent job explaining this.
I would think that as more satellites get launched they could use WCDMA and signal from your station could be seen by multiple satellites in orbit much like a cell phone can reach multiple towers.
Writing it out... I bet TDMA is required because the FCC would never grant a block of spectrum where hundreds of thousands of ground stations were using low gain, somewhat omnidirectional antennas to reach a constellation of satellites in space....
That is actually really cool.
So it is doing some combination of time division and spacial division.
I know Moore's Law is being repealed, but that's still how new types of electronics work, right?
Making small early batches of anything is more expensive per unit than making a ton of them —- regardless of whether it’s cutting edge tech or a plastic chair.
I assume that would be less expensive than creating larger circuit boards, shipping and packaging.
Starlink sats are low earth orbiting (about 250 miles away). The really high latency sats that people used to use were geosynchronous sats that are parked about 30,000 miles away, and the round trip delay between earth, bird, earth.
Its fantastic, can't wait until this is available where I live. Currently paying $200+ a month for 20Mbps from local wireless company.
2 miles North from here there is AT&T fiber and Comcast available, 5 miles South there is Comcast (150Mbps) but I'm in a small community of homes where only options are satellite or fixed wireless.
Since them most of these networks are still in place, now in the 100 mbit/gbit territory. And while most turned into commercial volcanoes over time, there are still some that work as community organizations to this day.
I wonder if it could improve gaming/video conferencing with people far away.
It suggests they could be lower latency than a great-circle path ground fiber without the satellite interlinks.
The idea of low orbit satellites for internet has been around at least since the late 1990's [0]