Elon Musk said SpaceX's multi 1000 satellite Internet service will be IP-less

https://twitter.com/elonmusk/status/967712110661615616

by deadcoder0904 • 8 years ago

132 102 8 years ago

sly010• 8 years ago

It's not "internet" then, it's a communications network. On top of which third parties will probably implement IP. He also says peer to peer, but perhaps he just means point-to-point, in which case it will be very similar to existing satellite communications network, perhaps just more and faster.

lallysingh• 8 years ago

Sounds like something analogous to ATM or whenever frames. The IP layer would probably start from the satellite transceiver on The customer premesis.

Numberwang• 8 years ago

For those of you in the know here. Is this a realistic project/enterprise he is working on?

Seems both costly to deploy that many satellites and risky in the sense that you can't count on getting a lot of customers. Especially with the new super speedy 5G networks and so on.

rkangel• 8 years ago

Even in the developed world, coverage of high speed wireless broadband (e.g. 3G and up) isn't complete. In the less developed world the coverage is even worse[1].

I don't know the costs to set up and run the network, but I do feel like there is a large possible customer base.

[1] https://opensignal.com/reports/2016/08/global-state-of-the-m...

mtgx2• 8 years ago

I think there was a recent report from the FCC that said only ~60% of Americans have access to 25+ Mbps internet. So, at least in theory, SpaceX has an enormous market.

swarnie_4• 8 years ago

Well populated and tech heavy parts of the UK have spotty and unreliable 4g which is baffling to me. If Elon can deliver a low latency satellite network even just for peering it will be a big step potentially.

derekp73• 8 years ago

I was at Motorola when Iridium went up, and witness first hand the changing market dynamics that were happening even before their first satellite was deployed. Was thinking then that if they mad the satellites generic data switches instead of a flying phone switch, they would have had a much bigger market.

So for Musk's network, it really depends on if 80% of the potential customers (people with money to spend on the service) get saturated with ground-level 3G - 5G by the time the network is operational or not.

Something else to consider, if the system allows for land-based carriers to tap into it and re-broadcast it on 5G (with different pricing). Imagine if there is no coverage in your area, and anyone can purchase a box to put on their roof that pulls down the satellite Internet and makes it available to everyone's 3G - 5G phones in the area. You lease the box from someone, and you get a commission on every client connection.

zaarn• 8 years ago

It would be most interesting indeed if the satellite network allowed operating your own ISP for extremely low cost. Making 3-5G available to phones and peering to existing networks maybe? (I'm not that familiar with Mobile ISP Politics, unlike BGP Politics)

Quarrelsome2• 8 years ago

it avoids all of the annoying issues that cartels spent a lot of time setting up on a state-by-state basis in the US (which results in limited choice).

Effectively you could connect to this anywhere in the world without the approval of local (e.g. Chinese citizens could get an unfiltered internet and the only "tell" is that they have a satellite dish). That makes the potential of this reasonably neat but it will depend on bottoming out the price and having enough capacity.

ec109685• 8 years ago

And governments not interfering with its operation.

geomark• 8 years ago

The equipment will no doubt be illegal to own in places like China. So it would get smuggled in or produced locally and sold on the black market. So far so good. But even if you install it surreptitiously I am pretty sure it will not be a LPI signal. So the government has a fleet of vans driving around sniffing for the spurious emissions and away you go to the gulag.

sho4• 8 years ago

Hell of a lot of places without 5G, or any, mobile internet. Don't forget aviation and to a lesser extent maritime either. People love their onboard wifi.

Owning the launch platform could mean that he's able to significantly undercut current players, although funnily enough soon-to-be competitor Iridium recently launched a whole bunch of new satellites using SpaceX. Good position to be in when your competitors are dependent on you!

There are certainly challenges, though. Latency is a huge problem with satellite internet, for obvious reasons. I'm not sure what the intended bandwidth is but satellite will always "feel" slower than mobile data of equivalent speed. The fact that it's a mesh network with who knows how many "hops" presumably makes it even worse. So it won't ever compete with mobile, not even 3G probably.

jelder• 8 years ago

StarLink will be mostly in LEO, so latency will be competitive with terrestrial services.

mrep3• 8 years ago

Don't forget aviation and to a lesser extent maritime either.

And trains, busses (it might be orders of magnitude cheaper than paying for cell data to provide those local hotspots people use on busses to/from work everyday), cars (I could easily see teslas using it over cell for data in the future), and rural areas where laying fiber is expensive.

jsjohnst6• 8 years ago

1) Starlink will be in LEO around 800mi up unlike most communications satellites (like HughesNet) which are in geostationary orbit at ~22,200mi. Expect 25-35ms latency as compared to 600+.

2) due to having a lot more birds in the air at lower altitude with newer tech, they should be competitive with 5G networks speed wise as compared to existing satellite internet.

Edit: fixed distance typo for HughesNet

castratikron• 8 years ago

For a while the Trump administration considered building a government 5G network to prevent Chinese companies from doing it first. American telecoms are dragging their feet so much that they are starting to pose a threat to national security. They say building a nationwide network could take three years, so I don't see much threat to SpaceX now.

https://www.wsj.com/articles/u-s-memo-urges-big-push-on-5g-w...

endymi0n• 8 years ago

Thinking at the scale of Musk's ambitions real quick, this makes a lot of sense.

At Tesla: Think about every Tesla car worldwide being able to phone home with self-driving learning data. Being able to remote operate them in the 0.001% of cases where human intervention is neccessary, no matter whether in the Australian outback or the Alps (without "no network").

At SpaceX: Being able to communicate with every Mars-bound spacecraft 24/7 by laser transmission, without having to rely on NASA's expensive Deep Space Network.

andygates• 8 years ago

The antennas for this are the size of pizza boxes - which puts a question over the car connectivity. That's a lot of flat sky-facing surface to hide.

Quarrelsome2• 8 years ago

How would the tech work exactly? Would this make it more difficult for the government to tap into connections as one could technically (outside of an ISP) jack straight into that network?

derekp73• 8 years ago

I would imagine that those governments would make it illegal for people to own the receiver equipment. Just like some (in the past) have made it illegal to own a shortwave radio.

john_minsk• 8 years ago

Agreed. Or Musk will have to agree to filter traffic.

deweller• 8 years ago

So the topography here is:

Individual router <-> satellite mesh net <-> SpaceX router connected to the internet.

Is there any reason why the protocol between satellites matters? They clearly aren't aiming for interoperability with other companies.

zitterbewegung• 8 years ago

Elon probably wants a mesh net so that SpaceX plans on satellites either failing or being able to add capacity at will. Elon designs his systems to be cheaper by making the systems more fault tolerant so it seems like that pattern would extend to this network. Also, he will be using extra capacity from SpaceX having more launch windows to put the satellites in orbit.

walrus015• 8 years ago

the protocol and modulation/coding scheme between satellites (for in-space RF links) matters a great deal, because coding efficiency can vary... If would have a great deal of protocol overhead and wasted bps if they were to optimized it to actually speak Ethernet between satellites (eg: to pass standard 1600B or 9000B frames) versus something that has been designed from the ground up for a very specific amount of latency, and designed for a satellite-to-satellite trunk link. However they're breaking up the data also has to deal with the possible earth station architecture. I have not personally seen the design of a Starlink satelilte, but the concept is basically similar to a very small version of o3b.

Each satellite will have basically three different RF setups on it:

1) L or S-band TT&C (tracking, telemetry and control). This is the admin and OOB management for the satellite and some of the frequencies can be seen in SpaceX's recent FCC filing, with earth stations in the Seattle and Los Angeles area.

2) Multiple Ka/V-band or other multiple spot beam antennas aimed straight down, serving CPEs.

3) There will be at least two dedicated high capacity dish antennas per satellite aimed at large, high capacity fixed earth stations (2.4 to 5 meter size) operating in the Ka-band or higher. The earth stations will need to be geographically distributed around the globe at various fiber IX points, similar to how o3b has done things now, so that the satellite-to-satellite trunk links can take the fewest possible number of hops before downlinking to the terrestrial internet. The more medium sized earth stations the better, because it increases the chance that a link to an individual CPE can take a bent-pipe path rather than traveling through multiple satellite-to-satellite hops. For example a CPE in far rural Wyoming, and a satellite passing over the general area of Wyoming, might be able to "see" a 3.0m sized high capacity earth station located near fiber in Cheyenne. Data from CPE, up to satellite, down to Cheyenne, one hop.

gagabity• 8 years ago

How will this actually work? Will individuals be able to tap into it skipping the local ISP? I also wonder how the issue of government censorship/control be dealt with in various countries around the world.

mtgx2• 8 years ago

It would be nice if it used unlicensed spectrum in all countries, and then all anyone would have to do to connect to the SpaceX internet is get their hands on a modem that supports it. That would be very liberating for many countries, and I'm not sure there would be much the local governments could do other than throw a tantrum. They could start requiring a license for all free spectrum, but that would have non-insignificant consequences for their local economy, too.

dboreham• 8 years ago

In general use of unlicensed spectrum for satellite communications is not permitted.

netsec_burn• 8 years ago

You forget that local governments can heavily fine SpaceX or, at worst, destroy their satellites.

jsjohnst6• 8 years ago

It would be nice if it used unlicensed spectrum

If you are talking about the bands I think you are referring too (2.4/5ghz) then you aren’t considering the implications. To send a signal from ~800mi away on those bands, you’ll need to be radiating with so much power that you’ll effectively destroy any terrestrial usage of those bands. Literally your Wi-Fi won’t work anywhere on earth due to the channel congestion / noise.

CamTin• 8 years ago

So is he not planning on offering Internet access then? Just transit to other Starlink stations? Presumably even that would allow somebody to offer a "real" ISP on top of this service with a ground station and then peering to the Internet. Still, paying for Starlink and then separately paying for a Starlink-Internet bridge isn't the greatest consumer experience.

madengr• 8 years ago

Assuming you have to pay for the latter. Similar to the way cellular companies give away free nanocells.

taneq• 8 years ago

Well presumably he'd have peering arrangements in a few key places, it just means that within the network your data would go relatively directly instead of satellite-hopping to a central server and then back out.

grondilu• 8 years ago

IP is fairly high in the TCP/IP layer stack[1], isn't it? So it shouldn't be too surprising that a transport layer (or link layer? Not quite sure) does not use it, should it?

1. https://en.wikipedia.org/wiki/Internet_protocol_suite

nsommer• 8 years ago

IP is a network layer protocol, it connects two hosts. TCP is a transport layer protocol, it connects two processes and runs on top of a network layer such as IP. The link layer connects two links (e.g. your computer and your WiFi router), examples are Ethernet or 802.11 (even though they also incorporate management of the physical layer).

What the satellites have to deal with is pysical/link level protocol to send packets from one to another (and from and to ground stations) as well as a network level protocol to route traffic inside the SpaceX network.

madez2• 8 years ago

I don't think the market for satellite Internet is only where other services don't provide fast enough internet. Of course, the technology of glass fiber in the ground can't be beaten qualitywise, but glass fiber can more easily be controlled by your ISP and your government, unlike satellite internet. If SpaceX manages to provide unfiltered, net neutral, cheap internet, I'd consider getting it to complement my current connection.

jsjohnst6• 8 years ago

but glass fiber can more easily be controlled by your ISP and your government, unlike satellite internet.

What would lead you to believe that? The reality is that the reverse is true, satellite internet is more heavily scrutinized as it’s far easier to intercept and very few companies are doing it.

madez2• 8 years ago

On my earthbound connection my government can influence the ISP, but how is the government of [insert-autocratic-country] here supposed to control what its citizen do over satellite internet?

I assume that all communication to, in, and from the satellite network is secured by cryptography to prevent eavesdropping.

zeristor• 8 years ago

Is this a case of doubling down, whilst SpaceX waits for the demand to launch satellites at the lower price grows.

Its one thing to be able to launch satellites far more cheaply than other companies, perhaps the economics demands certain number of launches a year to keep capacity up and price down?

The question is what happens when they launch all 1000 satellites?

swarnie_4• 8 years ago

I've always seen this as two things. The first being space x filling in gaps in their launch schedule as you've pointed out.

The second is to launch a commercial services which can provide a constant, predictable income to fund other projects. The upfront capital must be huge but with debt markets still this cheap i'm not surprised they're investing.

mkirklions2• 8 years ago

I always thought of it as Elon Musk advertising.

But I visit reddit where tesla/elon/spacex topics are boosted with upvotes.

Im just bitter because I believed in him, but his promises fell flat.

mrep3• 8 years ago

The question is what happens when they launch all 1000 satellites?

The plan is for 12,000 satellites and they are only designed to last a few years and then be replaced. This keeps launch rates constant, allows them to make much larger use of economies of scale by having that many satellites, and to constantly improve speed by building the new satellites with state of the art technology.

https://en.wikipedia.org/wiki/Starlink_(satellite_constellat...

zingmars• 8 years ago

That's a lot of satellites. What are they going to do with the deprecated ones?

imtringued• 8 years ago

The headline is completely misleading. The transport layer won't use IPv6. It's like saying ethernet doesn't use IPv6. That's unsurprising.

cornholio• 8 years ago

He specifically mentions a much lower protocol overhead, implying a translation of some sort, as opposed to just encapsulation. As an ex-programmer, is unlikely Musk would make that confusion.

arbitrage• 8 years ago

Are you ever an ex-programmer? I would say, once a programmer, always a programmer.

mozumder• 8 years ago

It's most likely a version of the ATM protocol[1]. Point-to-point, small packets, etc.. We used a version of that when I was working on the Iridium project.

Likely the exact same thing here, especially if they hired the same engineers.

[1] https://en.wikipedia.org/wiki/Asynchronous_Transfer_Mode

petee• 8 years ago

Beyond that, there is entirely no substance here - unless he wants to release details, this is non-news, just more business fluff - "Ours will be even better. It'll be huge. The best. Others, so sad"

"And if its hacked, we'll just patch the firmware. So easy."

kakarot• 8 years ago

I'd be bragging about my satellite fleet, too. The guy's only human.

mustardo• 8 years ago

Do you have small hands?

squarefoot• 8 years ago

Not to mention that "IP" may indicate not just Internet Protocol. My first reaction after reading the title was: "Wow! A satellite service using no patented hardware/software! Can't wait to find the project on Github!".

ksec• 8 years ago

The Media will pick it up, start to explain to average Joe that iPv6 is the latest tech, currently still only running x% of Internet. But Elon has something even better then IPv6, using encryption and Peer to Peer design!

I was going to say Stocks going up, then I realize SpaceX is not public yet.

I wonder if the Internet Services part will be sold to Telsa?

gnode3• 8 years ago

I wonder if the reason here is that the topology of the network will change rapidly. E.g. as equatorial and polar satellites cross each other, they will be in direct communication with each other for only brief periods of time. There are ways to cope with routing changes in IP, but I don't know if it'd be seamless enough.

westbywest2• 8 years ago

Comments up-thread speculate Musk may be looking at something like ATM, which uses small, fixed packet sizes for better flow control, along with expectation packets may arrive out of order due to traversing multiple paths. ATM has implementations for reordering packets, and the fixed packet size ensures reordering can be accomplished within a fixed latency.

KMag2• 8 years ago

It's easy to read too much into a tweet, but it sounds like the header size overhead and number of options are the primary reasons for not using IPv6.

Note also that IPv4 and IPv6 are nearly independent of the routing protocol.

Also, my first job was writing discrete event models of wireless communications equipment, including a couple satellite models. You'd be surprised how simple satellites themselves are. A lot of work goes into moving the complexity from the fragile, expensive, and out-of-reach satellites and into the ground stations. I wouldn't be surprised if packets weren't at least partially source-routed, which might very well involve broadcasting GPS-like tables of satellite orbit parameters.

gnode3• 8 years ago

I expect that internet traffic will still travel over the network using IP, but encapsulated in the SpaceX protocol. I expect translating the headers to and from IP would not make the protocol "simpler than IPv6", and would not be justifiable complexity to save a few bytes.

I think this "simple" protocol is just to simplify routing, and allow non-IP traffic over the network (in-band control messages; phone calls?).

qaq• 8 years ago

I think people are ignoring one very lucrative application Military.

mkirklions2• 8 years ago

People on here, you bet there are companies that lobby for this...

gnode3• 8 years ago

Not that I don't think it will be useful for this. But Iridium is already offering LEO satellite service for low-latency government / military use.

220V_USKettle• 8 years ago

IPX/SPX?

jgrahamc• 8 years ago

As long as they fix the "SPX window size can only be 1" it'll be fine.

rebelde• 8 years ago

I think Appletalk.

mAd-12• 8 years ago

Nahh, NetBEUI!

Non-routable and peer to peer, wcgw?

walrus015• 8 years ago

well, we can play Quake 1 across it...

sailfast• 8 years ago

I am excited for any option that would allow me to actually choose my provider as a consumer, rather than rely on embedded infrastructure. This is super exciting for that reason and many others (potential breaking national control of a global connection, for one) and I can’t wait to see how it progresses.

nugga• 8 years ago

If they can make it so you could get a cheap subscription to starlink, without needing your own receiver, to route your udp traffic I would pay for sure. Since it's going to be a LEO constellation the pings will be crazy low round the world and you could play US-EU games at under 100 ms even if you had to have a hop or two via fiber to the nearest starlink connected hub.

mentos• 8 years ago

I am excited about Space and self driving cars but I think this could have much greater positive effects on humanity. Freedom of the internet could be the thing that saves this planet when it comes to thinking on a thousand year term.

staltz• 8 years ago

Hi, I'm andrestaltz who Elon answered on Twitter. Like others have mentioned here, it seems like Elon is referring to the link layer between the satellites. "Peer-to-peer" probably meant "without a network switch".

I wish I had originally tweeted "... peer-to-peer connections within Starlink-only customers" because I was actually interested in knowing if Starlink could be like ZeroTier. https://zerotier.com/

deadcoder0904• 8 years ago

Sorry I'm the OP. I just stole the title from Reddit for some easy karma because I couldn't understand all the jargon. PS - Admire both your work <3

zerostar07• 8 years ago

what are the ping times for satellite networks?

mAd-12• 8 years ago

Horrible. I tested it on a plane for a client and we were getting between 1800ms to 2500ms at 28000feet.

gizmo385• 8 years ago

Depends on the network. The current satellite internet systems all use satellites parked in a geosynchronous orbit. This means that ping times are measured in the hundreds of milliseconds in most cases.

The satellite network that Musk is talking about uses a very large web of satellites in low Earth orbit, which would mean ping times of less than 100 milliseconds.

walrus015• 8 years ago

Musk is talking about the OSI layer 1 and layer 2 of the satellite network, which by definition needs to be something unique and custom. Nobody has ever built a LEO dual-satellite-CPE make-before-break architecture before. The closest thing is the o3b architecture which is intended for very large, costly customer terminals. This will still speak ipv4 and ipv6 just fine.

To gain a better understanding of why this needs to be totally unique/custom/proprietary, it is helpful to first have a thorough understanding of current high capacity dedicated geostationary orbit based systems (1:1 SCPC/MCPC with dedicated kHz) and various shared-bandwidth VSAT type systems (TDMA timesliced architecture between one large earth station, one piece of satellite transponder kHz, and a number of N fixed CPE terminals within the satellite's spot beam.

After understanding current geostationary architectures, dig into the "How" and "why" o3b was created and has been such a success, and its general system architecture which is proprietary.

Satellite engineer here: The OSI layer 1/2 needs to be totally custom because we're dealing with a unique architecture of, just off the top of my head:

a) dual satellite LEO architecture

b) possible satellite-to-earth station trunk links, and satellite-to-satellite

c) CPE terminals that have no moving parts and use phased array antenna systems to talk to two satellites at the same time. From the stationary point of view of a rooftop CPE, the satellite that is currently "rising" from the horizon and will be soon overhead, and the satellite that is currently overhead and will soon pass out of sight.

d) Densely packed high frequency spot beams on a moving LEO satellite. The closest thing that's ever been built to this before is again the o3b satellites, but there are a great deal fewer of them, they orbit much higher, and have much larger spot beams than these small LEO high-Ka/V-band satellites will have.

e) Custom indoor modem RF tech to talk to the rooftop CPE and provide a standard 100/1000 copper ethernet handoff (and possibly integrated 802.11ac dual band wifi). TDMA timeslicing per CPE and bandwidth allocation - there is no way that an individual CPE will get 1:1 dedicated bandwidth 24x7x365, the amount of capacity in an individual spot beam sized area will be oversold based on standard network architecture principles that most people don't try to max out the capacity of their circuit 24x7.

kartickv• 8 years ago

For people who're not technical engineers and without the knowledge to understand all the technical details:

- Does it make sense to make a phone that acts as a receiver (assuming the phone is outdoors), because it's one fewer device to buy and power, or does it bring down the cost significantly to have a fixed unit installed on your terrace?

- If we want to provide remote regions with Internet access, would you go with Starlink or O3b? Would your answer change if the requirement was affordable internet access, no matter the speed?

- Under what conditions can Starlink or O3b compete with terrestrial internet in cities?

Basically, I'm trying to understand the impact these programs might have on the world, not the technicalities.

walrus015• 8 years ago

This isn't going to be for phone sized things. it'll be more like the size of a direcTV dish but in a very different shape. The system design and technical requirements to have something that moves around and is handheld is very different than a fixed CPE that's aimed upwards.

For your second question, both are suitable, but in different markets. O3B is intended to replace high cost Ku and C-band transponder capacity for telecoms and ISPs that are in a place economically impossible to reach with PTP microwave or fiber. The smallest terrestrial o3b terminal is a pair of 1.8 meter motorized tracking antennas semi-permanently installed on concrete pads. O3B gives an ISP one big fat pipe back to the terrestrial internet in somewhere with good connectivity, and then that ISP can distribute service around their region using whatever technology they want to use, or have access to (PTP microwave, point to multipoint, fixed LTE, DOCIS3.0, various types of FTTH, etc).

o3b only functions at latitude +/-45. Starlink and similar systems will be in polar orbits, so with network architecture for satellite-to-satellite relay, there is the possibility of truly global coverage (as Iridium service works equally well at the north pole, in remote parts of Nunavut, at the south pole, in Tierra del Fuego, etc).

It will not be competitive in cities against terrestrial based infrastructure. Given advancements in how much data can be shoved through old copper (DOCSIS3/3.1, VDSL2 30a, g.fast) and various active ethernet FTTH and GPON FTTH, those will have much greater throughput. A dozen starlink satellites will have less aggregate data throughput than what it is possible to push down two strands of singlemode fiber.

devit• 8 years ago

Is there an estimate on the magnitude of the maximum total bandwidth achievable with a LEO satellite, both for a smartphone client and for a fixed home antenna?

How far would it be to 4G and gigabit fiber respectively?

patrickaljord• 8 years ago

It's not the bandwidth I'm curious about, I'm wondering what the latency will be like. Satellite connections usually have bad latency because of obvious physics laws limitation. This can be bad for real time apps. Still great for most web applications though.

simonvc• 8 years ago

These are MEO sats, not Geo stationary at 30,000km.

The round trip for these will be about London to Edinburgh and back and we video conference that distance all the time.

kamaal• 8 years ago

They were talking to Astronauts going to moon with resources way more smaller than this.

There is a saying that goes like "To build fast software, use a slow computer".

Elon Musk said SpaceX's multi 1000 satellite Internet service will be IP-less

Related Stories