Slashdot Mirror
The First Detailed Look at How Elon Musk's Space Internet Could Work (newscientist.com)
SpaceX has been granted permission by the Federal Communications Commission (FCC) to set up a vast network of thousands of low Earth orbit communication satellites. But the company has been tight-lipped about the project, known as Starlink. Mark Handley, University College London built a simulator based on public details from the FCC filings to understand the latency properties of the network. New Scientist reports: Although Musk has said he wants more than half of all internet traffic to go through Starlink -- Handley's simulation suggests that the project will be most appealing to high-frequency traders at big banks, who might be willing to fork out large sums for dedicated, faster connections. To create the simulation, Handley took what information he could from SpaceX's public FCC filings and combined this with his knowledge of computer networks. Initially, Starlink will consist of 4425 satellites orbiting between 1100 and 1300 kilometres up, a greater number of active satellites than are currently in orbit. There is only one way to arrange this many in a configuration that minimises collisions, says Handley. So he is confident that his simulation reflects what SpaceX is going for.
When sending an internet message via Starlink, a ground station will begin by using radio waves to talk to a satellite above it. Once in space, the message will be fired from satellite to satellite using lasers until it is above its destination. From there, it will be beamed down to the right ground station using radio waves again. Between distant places, this will allow messages to be sent about twice as fast as through the optical fibres on Earth that currently connect the internet, despite having to travel to space and back. This is because the speed of the signal in glass is slower than it is through space.
When sending an internet message via Starlink, a ground station will begin by using radio waves to talk to a satellite above it. Once in space, the message will be fired from satellite to satellite using lasers until it is above its destination. From there, it will be beamed down to the right ground station using radio waves again. Between distant places, this will allow messages to be sent about twice as fast as through the optical fibres on Earth that currently connect the internet, despite having to travel to space and back. This is because the speed of the signal in glass is slower than it is through space.
Just dropping by to mention that, while existing fiberoptic networks have index of refraction around 1.7 (so signal speed is c/1.7) , there is a relatively new thing referred to as "holey fiber." It's essentially analogous to microwave hollow guides, with the hole pattern sized to match the TE/TM modes of the injected light. The speed thru these waveguides is close to the vacuum limit.
https://app.box.com/WitthoftResume Code: https://github.com/cellocgw
High speed trading looking for a timing edge has already upgraded to shortwave which traverses an even shorter distance (bouncing through the lower atmosphere) right at the speed of light.
https://sniperinmahwah.wordpress.com/2018/05/07/shortwave-trading-part-i-the-west-chicago-tower-mystery/
SpaceX's revised FCC filing calls for about 1.6k of the initial 4.4k constellation to be at 550km orbit. Brings the minimum latency down to 15ms, instead of 25-35ms.
https://www.theverge.com/2018/...
https://forum.nasaspaceflight....
"'Tis great confidence in a friend to tell him your faults, greater to tell him his." --Poor Richard's Almanac
The lower latency could be useful for games, too.
Probably not, because the potential savings scale with distance. Halfway across the world and back is 40000 km/300000 km/s = 133 ms at light speed, at c/1.7 it's 226 ms so at most 93 ms to save. But I wouldn't try to play a twitch game at those ping rates while at reasonable gaming ping times the gains are mostly eaten up by the base latency. Unless you really desperately want to play with your guild from another continent, it's not going to matter much. It mainly matters for HFT where your buy/sell orders arrive a few milliseconds before the competition.
Live today, because you never know what tomorrow brings
Glass is 31% slower, not a rounding error. Speed of light through air is much closer to matching your description. While this article from 2013 talks about using air-based conduits, I don't think it's reached full deployment yet.
Fortunately, I have them.
Space Communications Protocol Specifications
Consultative Committee for Space Data Systems
CCSDS Technical Specifications
Space Assigned Numbers Authority
Spacecraft ID list and manual
Disruption Tolerant Networking
Exploration and Space Communications at NASA
Free Space Optical Communication
It's a small world and it smells funny; I'd buy another if it wasn't for the money; Take back what I paid (SoM)
I am a network designer, or network architect if you prefer. You are correct. A mesh satellite network which could beat a typical terrestrial link is not possible at this time. The only time Musk's current plan might compete on latency are for links across the pacific. Even then, often the terrestrial lines will win. LEO might be able to provide greater resistance to jitter, but there really is no way to predict that until typical traffic volumes, patters, and types are empirically provided as a test-load.
Currently the speed Musk insists his satellite network could provide is simply infeasible with current radio technology. He would have to find a way to break the laws of physics or make huge satellites with impressively focused and powerful wireless links in spectrum bandwidth which wont diffuse or be absorbed by the atmosphere. Currently there is no available spectrum in the volumes he would need.
There are many futurists here and elsewhere, with no background in networking. To try and convince them by any means I have found to be futile.
My firm has actually run the numbers. For this system to be profitable, the limitations in physics, assuming launch costs are free, the number of people needed based on current market tolerance for price at a given speed, the shared bandwidth will be slow DSL speeds at dial-up ping times. Amazing for distributed IoT. There is a market. But this will never be what Musk insists it will be so long as there is no revolution in our understanding of physics.
The FCC only has authority over the U.S. radio spectrum. Now he's got some 200 more approvals to go.
Iridium did the same thing on a smaller scale back in the 90's when the company was owned by Motorola . The tech was worked out very quickly. It was the politics that slowed them to a crawl. Motorola found out exactly how hard it was to get all the world governments and incumbent telcos to agree to give them a sliver of spectrum. They had initially planned on direct satellite to phone communications until several telcos raised their hand and said no way, they had to go through ground stations where they could listen in on the conversations as well as charge access fees.
Musk just fired Starlink's executive team because they couldn't meet his timeline. If they were bogged down trying to get spectrum, it's no surprise - it took Motorola years and thousands of meetings and bribes to pull it off. Motorola's final approval only came after a Motorola lobbyist waited until a specific country's representatives fell asleep. The lobbyist knew the reps were going to vote against approving the worldwide spectrum at an international telecom conference. The lobbyist delayed until 4am in the morning when very few representatives were present and the known going-to-vote-no guys were present but asleep. Only then did he bring Motorola's request up for a vote. It passed and Iridium was born
After all that drama, Motorola/Iridium ended up with a very skinny slice of spectrum. More spectrum means more bandwidth. Bandwidth is something Iridium isn't known for.