Elon musk free internet satellite

[tooki]

Unless Musk has found a way to massively increase the speed of light, then satellite internet will never compete with terrestrial internet simply due to latency. With modern web servers using HTTPS for everything, there’s a lot of handshaking going on for each HTTPS connection, so high latency kills web page loading performance, even if the actual bandwidth is high. (Same reason why 56K dialup and 56K ISDN had very different performance: ISDN had far lower latency.)

People wildly underestimate the importance of low latency for web browsing performance. It’s not just about gaming!

[rdl]

There's no way to know with certainty until this system is actually working, but I remember either Musk or SpaceX stating the additional latency will be almost unnoticeable since these satellite "constellations" will be in very low orbits.

[wraper]

Unless Musk has found a way to massively increase the speed of light, then satellite internet will never compete with terrestrial internet simply due to latency.

That's why those satellites will be at low altitude.

[BrianHG]

Unless Musk has found a way to massively increase the speed of light, then satellite internet will never compete with terrestrial internet simply due to latency. With modern web servers using HTTPS for everything, there’s a lot of handshaking going on for each HTTPS connection, so high latency kills web page loading performance, even if the actual bandwidth is high. (Same reason why 56K dialup and 56K ISDN had very different performance: ISDN had far lower latency.)

People wildly underestimate the importance of low latency for web browsing performance. It’s not just about gaming!

Did you think that Musk would get permission to place his satellites in the super-valuable geostationary orbit where you get that ~2 second round-trip light lag?

Also, the size of the antenna / reflectors for that distance with the necessary tranceiving power plus interference with those mass distribution TV satellites will hinder some established services worth big $$$.  You also get all the radio com signals mushed together at such a distance without a directional antenna on the ground.  Many smaller low orbit satellites can focus, as they orbit around, on small areas of land, even down to the size of a city.  At geostationary orbit, your lucky if you can focus your antenna beam diameter down to the size of a continent.

Remember, you are comparing something like 36 thousand kilometers to 0.2 thousand kilometers.  At 36000km, try to build a reflecting antenna to focus on a single city.  Then place 7000 satellites, each one never drifting, each focused on a different city sized point.

[madires]

LEO communication satellites aren't anything new. At the moment there are a few companies trying to set up networks based on that technology. And free internet for the world isn't a viable business plan.

[Rerouter]

Though I wonder if Datamining bandwidth limited internet with constant advertising redirects would be?

[rdl]

Interesting project. Probably has a better chance of succeeding than landing on Mars within a decade, but with an estimated total cost to develop and implement of US$10 billion, free internet isn't likely to happen.

https://en.wikipedia.org/wiki/Starlink_(satellite_constellation)

Apparently the current plan is for these satellites to orbit just above and below where the ISS is. The two prototype satellites are pretty big actually, at least I think so (400kg). It'll be interesting to see how small they can make production versions. If they're going to launch thousands of them they surely need to be cheap.

[beanflying]

The larger question is why are we discussing yet another twitter fart with zero details from Musk about something else he didn't invent but is claiming as the next great thing of HIS?

He truly has become the Edison of our time.

[Rerouter]

Discussing It because Not everyone out there will know why It will be hard to pull off if even possible, and Stealing some google hits away from copy-paste news sites to a technical forum is generally a benefit to humanity 

[beanflying]

Talk to some people who use the current Satellite and the latency with broadband in Australia with very limited user numbers makes that easy to figure out. Even going forward one or two generations of speed into the future (if they can be found) and it doesn't add up within 10-20 years.

Meanwhile the public keep getting told he is a visionary and look at the great ideas HE keeps having (Clearly he is a company of one). The media control and dodgy youtube accounts have corporate bs and positive spin written all over them and on top of that he receives free air time on mainstream media whenever he farts

Anyway enough Grinch typing 11.30pm Christmas Eve here. All the best to everyone even Musk 

[donotdespisethesnake]

No one has said it will be free, that is just nonsense.

Starlink satellites will operate at about 1200km, round-trip latency about 10 ms, which is completely different to geo-stationary (~35,000km).

If you doubt Musk can launch 12 satellites per week, you might have heard of Falcon 9, Falcon Heavy, or even BFR. There will be so much spare capacity he will be able to launch the satellites very cheaply. A lot of people seem to think Musk is a buffoon, and to be fair he says a lot of stuff that is very speculative if not daft, but what makes him quite unique is that he has a solid realistic business plan, and so far he has been able to deliver technically on those plans (if not on time).

[madires]

Besides some niche applications internet via satellite is complete nonsense. The data throughput of current submarine cables (fiber) is exceeding 100Tbit/s. But there's no limit for human stupidity.

[CatalinaWOW]

Terabits/second of transoceanic cable capacity does no good to me or the tens of millions of others who don't live within a couple of kilometers of a cable route.

Where this LEO solution might be of real use is solving what we call the last mile problem here in the US, and is presumably the last kilometer problem elsewhere.  An economical solution to getting high speed links to low population density areas has proven to be a tough problem.  How this costs out and performs will depend on the details of the implementation.   

And as others have pointed out it has a roughly 30:1 latency advantage over GEO solutions.

[wraper]

Besides some niche applications internet via satellite is complete nonsense. The data throughput of current submarine cables (fiber) is exceeding 100Tbit/s. But there's no limit for human stupidity.

You should move a little bit out of your comfort zone and find there is no or very poor internet in many places. Actually if you take all land area, only tiny part of it have decent internet.

[madires]

I know that very well. I had to deal with dial-in quite long. Still, LEO internet satellites can't solve the last mile problem for all the consumers, it's a niche application. Providing internet access to lots of consumers requires a massively scalable network design. Otherwise you end up with huge CAPEX and OPEX and also with rebuilding your network from scratch every few years while burning a lot of money. If you have telephone lines you deploy DSLAMs (up to about 50-250Mbit/s per subscriber), in case of cable TV CMTS' (up to about 400Mbits per subscriber), FTTH (GPON with up to 1Gbit/s per subscriber) or maybe 4G/5G mobile technology. They all have one very important thing in common: they are scalable. For more subscribers add more linecards or boxes, for more throughput add more fibers. If a new standard emerges replace old boxes with new ones. Back to the idea of LEO internet satellites. One satellite will handle maybe an aggregated throughput of 1Gbit/s, providing 10 subscribers with 100Mbit/s or 100 subscribers with 10Mbit/s downstream. How do we get the 1Gbit/s for each satellite to the correct satellite? Don't forget, LEO satellites won't stay in an area for too long. A: Lot's of uplink sites. And what about the subscriber? Will a US$30 box do the job? No, he needs a transceiver with antenna,. A bit expensive and not plug-n-play. And he will also have fun with rain, snow, fog, ice and what have you. How do you handle technology upgrades? You can't simply send a field engineer with a new box to a satellite. Deploying 8000 "Next Generation" satellites? Maybe you start understanding that the idea of internet satellites isn't feasible because of the lack of scalability.

[wraper]

They all have one very important thing in common: they are scalable. For more subscribers add more linecards or boxes, for more throughput add more fibers.

But they are not viable for remote areas and areas with scarce population. Expenses for building infrastructure would never pay off.

[Red Squirrel]

I think one way to make it feasible is to make it cost something, does not have to be a lot, but it should be something that you have to subscribe to, and without a subscription you can't access it. (would be encrypted etc)  This would limit abuse, once that's done, then it's the thing of allocating bandwidth, I think there should be a cheap (like $10/mo) package that is like maybe 100kbps.  There could be higher packages but then the cost goes up fast.  To me the idea of something like this is to act as a last resort/emergency way of getting internet and not as a primary way, as that does indeed seem unfeasible.  Trying to bring high speed globally is probably unfeasible but bringing in a low speed but semi usable connection may very well be feasible.

Though I do wonder if the cost is justifiable when you could put that money towards ground based fibre going to wireless towers.  Maybe a new wireless mesh protocol is even in order for this sort of thing.  You feed towers and then user devices or even other smaller towers can complete the mesh from the nearest tower via a high range but low speed wireless connection.  To gain access you'd then convert to ethernet or wifi. Make the protocol a standard and where anyone can participate, so they would put in the initial network, but anybody can buy the equipment and put up their own towers too and it would all be part of the same mesh. So imagine a lake of cottagers, they could chip in to put up their own tower and get line of sight to another tower, and so on.  Data would just take the best path, so if someone puts up another nearby tower and actually runs their own fibre then it will take that path.

There are mesh protocols like this but they arn't really widespread.  This would need to be plug and play where you can buy various gear to interface with it using wireless and wired.

[madires]

They all have one very important thing in common: they are scalable. For more subscribers add more linecards or boxes, for more throughput add more fibers.

But they are not viable for remote areas and areas with scarce population. Expenses for building infrastructure would never pay off.

Neither the deployment of any other technology. In the past this problem was solved by a state owned telco and the right to have a telephone, i.e. the cost of building and operating the network was shared. Today, with private telcos and their main focus on profit we need subsidies and regulation to get rural areas up to speed. There are also local initiatives, e.g. municipal FTTH networks, which are able to compete with commercial telcos easily. In most cases they provide more bang for the buck and some really cool features.

I'm aware that the big telcos in the US lobbied several states into passing laws denying municipals to build own networks.

[madires]

Though I do wonder if the cost is justifiable when you could put that money towards ground based fibre going to wireless towers.  Maybe a new wireless mesh protocol is even in order for this sort of thing.  You feed towers and then user devices or even other smaller towers can complete the mesh from the nearest tower via a high range but low speed wireless connection.  To gain access you'd then convert to ethernet or wifi. Make the protocol a standard and where anyone can participate, so they would put in the initial network, but anybody can buy the equipment and put up their own towers too and it would all be part of the same mesh. So imagine a lake of cottagers, they could chip in to put up their own tower and get line of sight to another tower, and so on.  Data would just take the best path, so if someone puts up another nearby tower and actually runs their own fibre then it will take that path.

We've tried that already with WLL, WMAN and IEEE 803.16. About 15 years ago there were several companies developing wireless meshed nodes - interesting stuff. But that technology never took off. Some of the ideas have survived in the form of projects like Freifunk.