Live data from Hacker News

SpaceX plans worldwide satellite Internet with low latency, gigabit speed

arstechnica.com

51–60 of 540 posts

Re: SpaceX plans worldwide satellite Internet with low latency, gigabit speed

#51

I don't understand the math. 800 satellites at 23 gbps seems to be able to provide 1 gbps to 18,400 connections at any given time. This seems to be many magnitudes off from what would be needed to have any impact. What am I missing?

It will be expensive bandwidth no matter what. The way they can make this work is if they 1) are targeting traditional satellite internet customers (ships on the sea, commercial aviation, etc) and 2) integration with 5th generation mobile networks trough wireless network virtualization. Mobile network provider can sell a service where traveling customer pays extra for connection that works in the middle of nowhere. W…

"commercial aircraft carriers"

That made me stop and think for a bit...

Re: SpaceX plans worldwide satellite Internet with low latency, gigabit speed

#52
post #22

I don't understand the math. 800 satellites at 23 gbps seems to be able to provide 1 gbps to 18,400 connections at any given time. This seems to be many magnitudes off from what would be needed to have any impact. What am I missing?

People aren't generally using a 1 gbps connection at 100% capacity 100% of the time. If we assume customers use an average of 1 TB of data per month and the time they use it is pretty spread out (seems reasonable since they are going to be spread out accross timezones eventually), you can serve about 6 million customers with 800 satelites @ 23 GBPS, and with 4400 @ 23 GBPS you could do ~33 million. Admittedly these n…

Still. What's the cost of deploying 800 sattelites vs cost of fiber infrastructure for 6 million people?

Re: SpaceX plans worldwide satellite Internet with low latency, gigabit speed

#53

And hella pingtimes ;)

> SpaceX expects its own latencies to be between 25 and 35ms

Unless you're doubting their math, that's not a high ping at all. In fact, that's a pretty amazing improvement given that existing satellite internet is >500ms. I didn't realize the difference between LEO and geostationary was that high.

Re: SpaceX plans worldwide satellite Internet with low latency, gigabit speed

#54
post #20

Earlier quoted context omitted.

Ok, given the election I have to ask.... Who's knowledge will we democratize?

Duh whatever Google and Facebook deem to be real news of course. The Ministry of Truth will root out all these evil white supremacist sites like Breitbart and Fox News. I long for the day when only Salon and Slate are considered sources of truth.

Risible tirades like these is why people don't take you seriously.

Re: SpaceX plans worldwide satellite Internet with low latency, gigabit speed

#55
post #51

Earlier quoted context omitted.

It will be expensive bandwidth no matter what. The way they can make this work is if they 1) are targeting traditional satellite internet customers (ships on the sea, commercial aviation, etc) and 2) integration with 5th generation mobile networks trough wireless network virtualization. Mobile network provider can sell a service where traveling customer pays extra for connection that works in the middle of nowhere. W…

"commercial aircraft carriers" That made me stop and think for a bit...

You mean like the Raft?

https://en.wikipedia.org/wiki/Snow_Crash

Re: SpaceX plans worldwide satellite Internet with low latency, gigabit speed

#56

I don't understand the math. 800 satellites at 23 gbps seems to be able to provide 1 gbps to 18,400 connections at any given time. This seems to be many magnitudes off from what would be needed to have any impact. What am I missing?

To take home users as an example, the residents generally sleep 8-10 hours a day, work 9 hours a day, commute ~1 hour a day, leaving only about 6 hours of free time at home a day. Assuming they spend that entire time watching 4K Netflix (25Mbit), they'll have used ~2TB in a month, or ~6Mbit/s total. Averaging bandwidth out over the whole month, each satellite would be able to support 3800 people. With the 4400 satellites they intend to deploy, they could support 16 million people.

Of course you generally need to provision for peak bandwidth rather than average so these numbers aren't even close to accurate but I think it's a rough explanation for why you don't need to fully provision every link.

Re: SpaceX plans worldwide satellite Internet with low latency, gigabit speed

#57
post #52
post #22

Earlier quoted context omitted.

People aren't generally using a 1 gbps connection at 100% capacity 100% of the time. If we assume customers use an average of 1 TB of data per month and the time they use it is pretty spread out (seems reasonable since they are going to be spread out accross timezones eventually), you can serve about 6 million customers with 800 satelites @ 23 GBPS, and with 4400 @ 23 GBPS you could do ~33 million. Admittedly these n…

Still. What's the cost of deploying 800 sattelites vs cost of fiber infrastructure for 6 million people?

It depends on how far apart those 6 million people are and how much infrastructure is already in place.

Re: SpaceX plans worldwide satellite Internet with low latency, gigabit speed

#58
post #32

How do you get an uplink? I generally thought of satellites as providing pretty good downlink, but high latency because the uplink was through some alternative network - eg. phone/dsl lines.

It depends on the satellite. The higher they are, the higher the latency / lower power / lower effective speed. Spacex plans for very low orbit. They're going for 1,110-1,325 km heights, which should give them the planned Existing Iridium is at 781 km and offers rather slow and expensive, global connections. But they provide phones, with reasonable-sized antennas: https://en.wikipedia.org/wiki/Iridium_Communications

Re: SpaceX plans worldwide satellite Internet with low latency, gigabit speed

#60
post #27

I don't know how they're going to get 25/35ms hops. If they're using low orbits to reduce latency, the surface coverage is going to be lower so they're going to have to mesh them in some way to span large areas. That means routers and intelligence which means uplink, interlink, downlink hops in both directions. 600ms here we come again.

Here is the part you missed: "Optical inter-satellite links permit flexible routing of traffic on-orbit."
Post reply on HN