Vint Cerf Is Working on an Internet for Outer Space (quantamagazine.org) 86
"TCP/IP doesn't work at interplanetary distances," 77-year-old Vinton Cerf tells Quanta magazine. "So we designed a set of protocols that do." Specifically, bundle protocols: a disruption/delay-tolerant networking (DTN) protocol with nodes that can also store information:
A data packet traveling from Earth to Jupiter might, for example, go through a relay on Mars, Cerf explained. However, when the packet arrives at the relay, some 40 million miles into the 400-million-mile journey, Mars may not be oriented properly to send the packet on to Jupiter. "Why throw the information away, instead of hanging on to it until Jupiter shows up?" Cerf said. This store-and-forward feature allows bundles to navigate toward their destinations one hop at a time, despite large disruptions and delays...
So, a couple decades after conceiving of bundle protocols, is the interplanetary internet up and running?
We don't have to build the whole thing and then hope somebody uses it. We sought to get standards in place, as we have for the internet; offer those standards freely; and then achieve interoperability so that the various spacefaring nations could help each other. We're taking the next obvious step for multi-mission infrastructure: designing the capability for an interplanetary backbone network. You build what's needed for the next mission. As spacecraft get built and deployed, they carry the standard protocols that become part of the interplanetary backbone. Then, when they finish their primary scientific mission, they get repurposed as nodes in the backbone network. We accrete an interplanetary backbone over time.
In 2004, the Mars rovers were supposed to transmit data back to Earth directly through the deep space network — three big 70-meter antennas in Australia, Spain and California. However, the channel's available data rate was 28 kilobits per second, which isn't much. When they turned the radios on, they overheated. They had to back off, which meant less data would come back. That made the scientists grumpy. One of the JPL engineers used prototype software — this is so cool! — to reprogram the rovers and orbiters from hundreds of millions of miles away. We built a small store-and-forward interplanetary internet with essentially three nodes: the rovers on the surface of Mars, the orbiters and the deep space network on Earth. That's been running ever since.
We've been refining the design of those protocols, implementing and testing them. The latest protocols are running back-and-forth relays between Earth and the International Space Station... We did another test at the ISS where the astronauts were controlling a little robot vehicle in Germany.
So, a couple decades after conceiving of bundle protocols, is the interplanetary internet up and running?
We don't have to build the whole thing and then hope somebody uses it. We sought to get standards in place, as we have for the internet; offer those standards freely; and then achieve interoperability so that the various spacefaring nations could help each other. We're taking the next obvious step for multi-mission infrastructure: designing the capability for an interplanetary backbone network. You build what's needed for the next mission. As spacecraft get built and deployed, they carry the standard protocols that become part of the interplanetary backbone. Then, when they finish their primary scientific mission, they get repurposed as nodes in the backbone network. We accrete an interplanetary backbone over time.
In 2004, the Mars rovers were supposed to transmit data back to Earth directly through the deep space network — three big 70-meter antennas in Australia, Spain and California. However, the channel's available data rate was 28 kilobits per second, which isn't much. When they turned the radios on, they overheated. They had to back off, which meant less data would come back. That made the scientists grumpy. One of the JPL engineers used prototype software — this is so cool! — to reprogram the rovers and orbiters from hundreds of millions of miles away. We built a small store-and-forward interplanetary internet with essentially three nodes: the rovers on the surface of Mars, the orbiters and the deep space network on Earth. That's been running ever since.
We've been refining the design of those protocols, implementing and testing them. The latest protocols are running back-and-forth relays between Earth and the International Space Station... We did another test at the ISS where the astronauts were controlling a little robot vehicle in Germany.
Far out! (Score:2, Funny)
Re: (Score:2)
Too far out.
"A data packet traveling from Earth to Jupiter might, for example, go through a relay on Mars, Cerf explained. "
Why in the world would you send a packet to Jupiter via Mars?? Does Cerf think that the planets are neatly lined up in a row, like on that poster in grade school, and Mars is on the way to get to Jupiter?
Spacecraft have tiny little dishes, compared to Earth dishes. MRO, with the largest com antenna of any orbiter ever sent to Mars, has a 3 meter-diameter antenna. Compare that for a m
Re:Far out! (Score:4, Insightful)
Why in the world would you send a packet to Jupiter via Mars?? Does Cerf think that the planets are neatly lined up in a row, like on that poster in grade school, and Mars is on the way to get to Jupiter?
Because Jupiter might be on the other side of the Sun and can't be reached from Earth.
However, it would probably be smarter to have relay satellites in earth's orbit, offset by 90 degrees. This way you would always know where the relay is going to be in regards to earth.
Wait the week, it makes more sense [Re:Far out!] (Score:3)
Why in the world would you send a packet to Jupiter via Mars?? Does Cerf think that the planets are neatly lined up in a row, like on that poster in grade school, and Mars is on the way to get to Jupiter?
Because Jupiter might be on the other side of the Sun and can't be reached from Earth.
He wants them to design a whole com protocol around solar conjunction, an event that lasts for about a week every 13 months, with a solution that degrades com rate by a factor of something like (70/3)^2= 540, and means you have to repurpose an asset in orbit around Mars which has other science tasks to accomplish?
Just wait the week and send the data back at high speed.
However, it would probably be smarter to have relay satellites in earth's orbit, offset by 90 degrees.
Or, better yet, 60 degrees.
Re: (Score:2)
>Or, better yet, 60 degrees.
Is there some advantage to sticking your critical relay satellite in the middle of a chaotic asteroid field? 90 degrees is a less long-term stable orbit, but it is clear of the asteroid fields, and you probably want to do some station-keeping regardless.
Space is big [Re:Wait the week, it makes more...] (Score:2)
“Space is big. You just won't believe how vastly, hugely, mind-bogglingly big it is. I mean, you may think it's a long way down the road to the chemist's, but that's just peanuts to space.”
-- Douglas Adams, The Hitchhiker's Guide to the Galaxy
There could be an "asteroid field" in the Earth trojan points, but so far there's only one known Earth Trojan, 2010 TK7. Even if there were billions of them, though... the libration orbits around the Earth-sun Lagrange point span a volume of space
Re: (Score:2)
Or am I confusing the number of searches with the number of results? Whatever - the number of Earth Trojans is far lower than was expected when people started looking. It looks as if the relatively close, relatively frequent passes of Mars and Venus are sufficien
Re: (Score:2)
Those are the big ones, floating hills and mountains observable with oblique lighting from ~150 million km away. They'd totally obliterate any satellite in an impact and barely notice.
Now, want to venture a guess as to how many free-floating pebbles are present in the same space? Any one of them would likely also do catastrophic damage on impact.
Re: (Score:2)
I haven't heard of any such IR searches focussed on the Jovian trojan system, but it may be on the agenda for the next big space
Re: (Score:2)
Big asteroids are relatively rare. Gravel potentially much less so, and almost as dangerous at orbital speeds.
In the Belt at least (we really don't have much detailed data about Earth's Trojans), the number of asteroids of a given size very roughly increases tenfold with every halving of diameter. There's an estimated 25 million Belt asteroids 100m across, if the size distribution continues down to gravel sizes, then at a wild estimate there's over 250 quadrillion fist-sized asteroids, and half a sextill
Re: (Score:2)
He wants them to design a whole com protocol around solar conjunction, an event that lasts for about a week every 13 months
Or about every month for Mercury and 6 months for Venus.
Re: (Score:2)
I'm thinking it's fido net again, just on a solar system scale.
I'll take the moon hub, because I can hit Venus or Mercury
most of the year and if not, relay it via Mars.
3 days for email, not to shabby.
Re: (Score:2)
Re: (Score:2)
Why in the world would you send a packet to Jupiter via Mars?? Does Cerf think that the planets are neatly lined up in a row, like on that poster in grade school, and Mars is on the way to get to Jupiter?
Yes, that must be it. Clearly Vint Cerf isn't the giant in the field that we thought he was and you Geoffrey Landis, Slashdot poster 926948 are the true genius.
Joking aside, Cerf is just providing an example here. The example is of a perfectly credible scenario. Space networks are a very interesting area, really. You have massive amounts of data that you want to send, and tiny, unreliable pipes. So, what data you send has to be very carefully selected and prioritized. I'm quite sure a lot of consideration a
A silly example [Re:Far out!] (Score:2)
Why in the world would you send a packet to Jupiter via Mars?? Does Cerf think that the planets are neatly lined up in a row, like on that poster in grade school, and Mars is on the way to get to Jupiter?
Yes, that must be it. Clearly Vint Cerf isn't the giant in the field that we thought he was
Remind me again, which planetary missions was he on?
Joking aside, I'm sure that he was simply straining to come up with quick example suitable for a semi-popular-magazine interview that non-experts would be able to instantly understand. It's just, the example he picked was a silly one.
and you Geoffrey Landis, Slashdot poster 926948 are the true genius. Joking aside, Cerf is just providing an example here. The example is of a perfectly credible scenario. Space networks are a very interesting area, really. You have massive amounts of data that you want to send, and tiny, unreliable pipes. So, what data you send has to be very carefully selected and prioritized. I'm quite sure a lot of consideration and modelling went into designing these protocols.
And you do know that we do that already, right? Every Mars orbiter since the late lamented Mars Observer has had data relay capability, we've been using it routinely since MER (since Pathfinder, if you include the rover to la
Re: (Score:2)
Joking aside, I'm sure that he was simply straining to come up with quick example suitable for a semi-popular-magazine interview that non-experts would be able to instantly understand. It's just, the example he picked was a silly one.
Well, I agree that he was just coming up with a quick example. I think we disagree on how silly it was. There are plenty of planetary configurations where it makes sense to relay through Mars or some other intermediate waypoint rather than transmitting directly to Jupiter. There are a number of reasons, but the main one (aside from being blocked by the Sun) is path loss due to distance. You can transmit more data in less time if you use a relay. This might not make a difference if your network is overflowin
Re: (Score:2)
And we've had that relay capability because the problem was recognised, and Cerf (and others) brought in to engineer a long-term, updatable, scalable solution. Which is now being deployed.
Re: (Score:2)
Because at times Mars is "on the way" to Jupiter. Just as on Earth, the protocol has to be capable of self-adapting to changing conditions. When it comes to communications, space is even more about time than about distance.
There's this thing called "the sun" (Score:2)
There's this thing called "the sun" that puts out a little bit of electromagnetic radiation. Sometimes it hangs out between Earth and Jupiter.
> Compare that for a moment to the 70 meter Deep Space Network antenna at Goldstone
Yes, try pointing that at the sun. And trying to pick up a probe on the other side of the sun and a little to the left.
UUCP ftw (Score:4, Insightful)
Re: (Score:1)
Re: (Score:3)
You'd need a whole new kind of UUCP map to figure out what was going to be close to what when, and how long it was going to take to send a signal at that time, etc, in order to properly route messages. You might reasonably bang out paths if you wanted to do the calculations yourself on what the fastest way to send your message was, but ideally the systems themselves would figure it out.
Not ready? Just wait longer. (Score:2)
"So, a couple decades after conceiving of bundle protocols, is the interplanetary internet up and running?"
So, a couple decades after conceiving newer improved protocols, is the IPv6 internet up and running?
Yeah, exactly. Good luck with that whole interplanetary gimmick. I'm sure we'll have at least a dozen mega-corps running countries by then, each fighting for years to be recognized as THE golden protocol standard.
Like this? (Score:2)
https://www.nasa.gov/content/d... [nasa.gov]
So... UUCP? (Score:4, Interesting)
If you used UDP, had a gateway to convert packets to something else that could be transferred by UUCP, and used an efficient protocol for transfers... you'd have your store-and-forward system.
No doubt it would need some architectural work along the way, and you'd need some kind of much faster store for the packet data...
I thought about this back when UUCP was still a thing, and space stations weren't. A friend of mine used to say that eventually our microwave ovens would contain full operating systems simply because it would be cheaper. We're not there yet, but I'd say we're about half way. You see AVR chips where a relatively simple discrete circuit would do the job (in a furnace control board, where I just saw an attiny a few days ago, for example) because they simplify, and thus reduce development time and cost. How long before attinys are replaced by more complex SoCs? They already have more CPU power than the moon lander.
Re: So... UUCP? (Score:2)
Re: (Score:1)
Also because CPU systems are cheaper in production, not just development. An 8-bit CPU where all it does is waggle a single GPIO pin up and down is cheaper and needs less board than a 555 and the three discretes it needs. If you need something a little more complicated like debouncing a push button, shoving a CPU in there is a complete no-brainer. As a bonus, you get to do three switches at once with an 8-pin device at no additional cost.
Re: So... UUCP? (Score:2)
The moment the device needs to access the Internet. Then, you'd use something like an ESP32 or ESP8266 (which is basically a wi-fi chip with a second core, some extra ram & i/o capabilities so you can add it for the wi-fi, and get general microcontroller capabilities as a free bonus). Once the ESP8266 came out, it was "game over" for 8 bit MCUs because it's cheaper to just enhance a wifi chip slightly to do double-duty than to bother with trying
sudo apt-get install uucp (Score:1)
Re: (Score:2)
Re: (Score:2)
Re: (Score:1)
Have you guys ever seen a bullet chess tournment? E.g.: https://www.youtube.com/result... [youtube.com]
Glad I Saved Those Earthlink Disks (Score:1)
Who knew they would be useful again?
In space no one can hear your modem.
Woo & Hoo (Score:2)
Quantum Entanglement (Score:2)
Wasnt quantum entanglement supposed to solve the lag issue? By coding a single onto an atom that is entangled, it was supposed to be viewed simultaneously on its entangled twin regardless of how far apart.
Re: (Score:2)
By coding a single onto an atom that is entangled,
That's still the best way. You buy the whole album encoded onto a molecule and there's usually only a couple good songs anyway.
Re: Quantum Entanglement (Score:3)
You need two bits of classical information transfered for every entangled state if you want to receive anything else than perfect random noise.
Physics is a bitch.
Re: (Score:3)
Wasnt quantum entanglement supposed to solve the lag issue?
Nope sorry. It doesn't work like that.
By coding a single onto an atom that is entangled, it was supposed to be viewed simultaneously on its entangled twin regardless of how far apart.
You'd have to send the atom. And that would be slower than just sending the photons instead.
Re: (Score:2)
You'd have to send the atom. And that would be slower than just sending the photons instead.
Setting aside that you can entangle photons.
The "teleportation" quantum entanglement communication concept is to realise that you can separate the transmission of the medium and the transmission of the data.
Essentially you entangle and send two constant streams of photons, one stream goes to the destination, one stream goes into an equal sized buffer. This transmission has all the classical constraints around time. This transmission also doesn't contain any data.
You then encode the data onto the buffer str
Re: (Score:2)
Setting aside that you can entangle photons.
Why go to all the trouble of sending entangled photons when you can just trivially send photons instead? How is that going to be any faster...
Remember we are talking about lag here, not any kind of cryptography or anything. Just speed.
The "teleportation" quantum entanglement communication concept is to realise that you can separate the transmission of the medium and the transmission of the data.
Yes. But with photons what's the point it's not any faster. And with atoms....
Remember again, lag. Not any kind of cryptography or anything. Just speed. Transmitting any kind of medium is going to be slower than the photons. Or they are going to just be photons. Use those.
Essentially you entangle and send two constant streams of photons, one stream goes to the destination, one stream goes into an equal sized buffer. This transmission has all the classical constraints around time. This transmission also doesn't contain any data.
Wh
Re:Quantum Entanglement (Score:5, Informative)
Unfortunately, no. Entanglement has some interesting properties, but it can't send classical information fast than light. This is to the great annoyance of many physicists, engineers and stock traders, but such are the rules of the universe.
You can still use entanglement for other things though - most usefully, key-negotiation to establish provably-secure, completely-impossible-to-intercept communications.
Re: (Score:1)
Until someone develops the Atom in the Middle attack.
Re: (Score:2)
The physics works out as completely impossible. That's why it's of interest. There may still be implementation vulnerabilities, like sensors susceptible to saturation if an over-intense signal is injected.
Re: (Score:2)
You can still use entanglement for other things though - most usefully, key-negotiation to establish provably-secure, completely-impossible-to-intercept communications.
Except that in practice mathematical cryptography is so much easier to get right that quantum cryptography is both more expensive and less secure.
Re: (Score:2)
There's no provably secure asymmetric encryption algorithm though. We only have practically secure ones - they can all be broken by sufficient processing power, though the amount of processing power required on average may be on the the 'use the entire output of the sun to power computers until it burns out' type of scale. They are all also subject to the small risk of a breakthrough - that slim chance that, somewhere in the world, a math enthusiast is going to come up with some key insight that no-one has
Re: (Score:2)
There's no provably secure asymmetric encryption algorithm though.
True, but I'm not seeing the relevance to quantum crypto, which isn't really an encryption technique at all; it's a key distribution technique whose resulting key bits are used in a symmetric algorithm (typically XOR).
We only have practically secure ones - they can all be broken by sufficient processing power, though the amount of processing power required on average may be on the the 'use the entire output of the sun to power computers until it burns out' type of scale. They are all also subject to the small risk of a breakthrough - that slim chance that, somewhere in the world, a math enthusiast is going to come up with some key insight that no-one has had before and figure out a quick and easy way to factorise the products of very large prime numbers, or something similar.
Unless we can prove that P != NP, in which case trapdoor functions based on proven NP-hard problems (which factorization isn't, and isn't thought to be, though DLP is proven to be) won't be vulnerable to those sorts of insights. They may still be vulnerable to quantum computers, though. I'm no
Re: (Score:2)
I'm not sure what the NP-hard status of the various quantum-resistant asymmetric algorithms is.
Some are reducible to known NP-hard problems, some aren't: https://en.wikipedia.org/wiki/... [wikipedia.org]
This is heroic (Score:2)
Re: (Score:2)
Those Mars colonists will need all the porn they can get.
More importantly. someone needs to make sure they are getting tracked all the time and delivered advertisements at every opportunity.
Re: (Score:1)
It's not IPFS. (Score:2)
I was hoping more recognition for IPFS. It's a really nicely designed architecture, and in need of more development and deployment.
Re: (Score:2)
Ditto. I've actually used a demo app with IPFS. Wondering what problem is Vint trying to solve that hasn't already? Hardware?
Re: (Score:2)
The story is seriously lacking in detail on that, but if you look around a bit the answers are elsewhere: Bundle protocols. The fundamental idea is truly ancient - nodes that store entire messages, forwarding them as unreliable links become available. The same principle that once ran UUCP and bang-path email back in 1980. Such methods of communication fall into obsolescence with the rise of reliable global networking, but modern protocols based on the same concepts are potentially useful in space communicat
Re: It's not IPFS. (Score:2)
Yeah. You mean buffering. Just an extraordinary timeframe. Measured in milliseconds, feh. In 10s of seconds, annoying and file killing. In light-minutes, paradigm shattering. In multiple revolutions, 'gee, we need a new protocol'...
Re: (Score:2)
Buffering combined with message routing. It's not new, it just needs turning into a formalised and proven protocol.
Sounds like packet radio (Score:2)
IIRC Packet radio is built on a similar store-and-forward mechanism.
Re: (Score:2)
So was/is FIDO.
Re: (Score:2)
Half-true. Packet radio itsself doesn't use store-and-forward, but some of the protocols that historically ran on top of it were, because long-range packet radio communications are not reliable and links had to be used opportunistically when the fickle gods of the ionosphere were feeling favorable.
Store and Forward... (Score:1)
Re: (Score:2)
Pluto is still only 5.5 Light Minutes away
??? I think you're off by a factor of about 50 there. How did you get 5.5 minutes to Pluto when you have about an hour to Jupiter?
Also, I think you're dismissing the challenges of an interplanetary network out of hand a little too easily. This has been designed for what exists now and the near future. Also, you have no way to prove that you can forward a packet without losing it until you have actually forwarded the packet without losing it (i.e. you get a confirmation). Until you have massive amounts of b
Re: Store and Forward... (Score:1)
Re: (Score:2)
Sorry, are you talking about a typo on my part or your part? Do you mean the number of minutes to get a signal to Pluto?
Re: (Score:1)
Re: (Score:2)
Ah, 5.5 hours. I actually did figure it out, but in terms of minutes since the other terms were in minutes. I came up with about 285 minutes, which is only 4.75 hours, so I didn't recognize the 5.5 figure as being in hours. Part of the problem here is that Pluto has a more elliptical orbit than the other planets and it is also out of the orbital plane of the rest of the planets. So the distance to Pluto I was relying on to come up with 285 minutes was the current distance. Otherwise, it is between 238 (~4 h
Re: Store and Forward... (Score:2)
I think that by the time we have more than a thousand people living on Mars "for real", we'll have a ring-topology network around the Earth's orbit with two (probably laser based) paths between Earth & Mars, and enough bulk bandwidth to use lots of forward error correction. The chokepoint won't be between Mars-orbit and Earth-orbit, it'll be between Earth-orbit and Earth-surface. Between orbits, latency will always be a bitch, but scaling up to bulk bandwith between them should be fairly easy... at leas
Re: (Score:2)
>Doesn't make a lick of sense once you have an orbital sattelite network like say StarLink and "MarsLink" that kicks "orientation" right out as there will always be a dedicated orbital satellite pointed in the direction you want to send or receive.
Umm, no? Every Starlink satellite is going to be pointed at Earth. Some will be pointed in the right direction to send a signal to Mars, but Earth will be in the way, and the satellites are only a few hundred km up, Earth fills their entire field of view. Be
Re: (Score:1)
You would not point the LEO starlink satellites at mars..
Re: (Score:2)
>once you have an orbital sattelite network like say StarLink and "MarsLink" that kicks "orientation" right out as there will always be a dedicated orbital satellite pointed in the direction you want to send or receive
How do you read that to NOT suggest that it's an element of Starlink doing the work?
I agree, interplanetary relay satellites make a lot more sense, but then why mention Starlink at all? Planetary and interplanetary communication satellites have absolutely nothing to do with each other. Di
Re: (Score:1)
Elon Musk's companies are all technology flywheels... the innovations at one feed into the next and one project to solve a problem solves problems in other areas.
StarLink has trialed laser based communications between satellites to conserve radio spectrum, but guess what that can also be used for... interplanetary links. If you are using lasers you may not even have to modify the satellite significantly, you just make a variant of a normal starlink satellite with more laser links and no r
Re: (Score:2)
Doesn't matter if you're using lasers or radio, the signal will still be 360 billion times fainter at Mars than at an "adjacent" satellite (I mis-typed in my first reply). Lasers can offer pretty much the minimum theoretical signal dispersal, but they still obey the inverse square law.
To establish a link across interplanetary distances you're going to need completely different transceiver hardware, and a completely different communication protocol since TCP/IP requires timely confirmation. Sure, you could
Re: (Score:1)
Good news to start the week. (Score:1)
Submarines? (Score:2)
It seems something like this might also work well for getting data to (and maybe from) nuclear submarines deep under the ocean. Granted, I could be way off base on this, but I know that is a challenge for numerous reasons.
Internet for space ....... (Score:2)
Latency sucks.... no playing Stadia from the Martian Colony. Never underestimate the throughput of a Spacex Starship packed with NVme Drives.
timestamps? (Score:2)
What do you use for your timestamp? Earth UTC?
Need to read the spec one of these days.
Nice (Score:2)
c:\>ping 237.122.122 //mars substation
Reply from 237.122.122: bytes=32 time=324456743 hours, 2 minutes, 5 seconds, 22 ms TTL=64
Re: (Score:2)
Re: (Score:2)
I'm more concerned that the TTL for that ping is apparently 64 decades.
I knew US Internet was bad. But that's like creating an IP Datagram Over Snail (IPDOP) extension to RFC 1149.
Failed at IPv6, let's try something really useless (Score:2)
Vint Cerf is the mastermind behind the IPv6 debacle. So yeah, kick him upstairs, he's done enough.