Vint Cerf’s Plan for Building An Internet in Space
quantamagazine.orgWhat about neutrino communication? It's the biggest spectrum we haven't analyzed yet and it would perfectly make sense for an advanced civilization because it pass through matter much better than any other alternative.
Passing through matter isn't a big problem with interplanetary communication? With a few antennas distributed around the respective planets (and a few in orbit if you want to work during eclipses) you can ensure line-of-sight transmission.
Did anyone do High Frequency Trading with neutrinos yet?
no, but i heard tachyons are great for dealing in futures...
I've always wondered about the feasibility of using space for temporary/short-term storage needs.
I remember a time when, as a junior developer, I'd log into operations, and hand the operator my tape reel with my work for the day - and she'd go off, mount the spool, and start the streaming, until my files were available in local storage for a few hours, and then .. when logging out, I'd have the operator re-stream the files, and off we go ..
So, why not use space as a large tape buffer? Send the satellites off, 20 minutes of light-speed away, and use the big gap as a cache.
Seems sorta feasible to me, but I guess I'm overlooking some ridiculously obvious factor, such as launch costs, viability, etc. But really, seems like a fleet of satellites spread out around the solar system could function as quite a nice medium ..
Probably the same reason that you don't store drinking glasses by throwing them really high up in the air.
See pingfs. It stores data on ICMP packets.
What is old is new again (just with a much bigger buffer):
When there's a 40 (mars) or 80 (jupiter) minute delay that's guaranteed, it sounds like a unique form of a time lock. Put your private pgp key in a 24 hour time delay circuit and guarantee that no one can do a decryption in the interim (long as it's the only copy).
I wonder if 50 years from now there's going to be a blockchain based on that technology (as opposed to the computation or storage based blockchains we have now).
I think the thing that you're missing is that there's no shortage of physical space right here? Said differently, why not build that satellite, load it up, and instead of launching it.... put it in the closet. Get it out when you need it. What advantages does that launch have?
GP doesn't want to store the data on the satellite, but rather use the space in-between as a Friday line storage. The satellite would be a repeater that retransmits what it receives. If it was 20 light minutes away, there would be 40 minutes worth of data in transit, so the whole system could store 40 minutes long circular buffer.
Might be a bit lossy :)
Right, so you have to send CRCs as well.
The nice thing about this scheme is that you don't technically need to send an active repeater: the mirrors on the moon could already serve this purpose.
At the end of the day, it's a tradeoff between the cost of extra storage, the sequential aspect of data retrieval from a delay line, and its energy cost.
I believe he's implying the use of the satellite as a repeater and the intervening space as a delay line
I do not think it's viable in any sense with launch costs but in a hypothetical scenario where you had something 20 light minutes away with a channel that had no better use you certainly could buffer 40 minutes of data on the channel
It is limited here not by the availability of space but the ability to collimate the signal and the availability of power density at the receiver to discern the signal from noise
Yes indeed, I imagine a situation where some natural resource is modulated such that, 20 minutes later here on Earth, 1000's of Petabytes of data is received such that, if re-transmitted, it will be reliably received across the distance of space. Lets say 20minutes, 1 year, 10 year and ... 1000 year intervals .. might be the line item deals on offer.
Truly a sci-fi concept, inasmuch as I have no idea of the science involved, but I can still imagine some scenario where such valuable data becomes committed to eternal space.
Okay, so let's say you get a really good signal between your satellites and manage 1mbps at interplanetary distances.
Even if you get a super cheap launch you just spent tens of millions of dollars on storing... 300 megabytes.
.. where nobody else can access it.
The protocols are interesting, but it seems to me that point-to-point is more applicable in the vast emptiness that is outer space. Why send a packet to Mars when Jupiter is 5° to the left and I have a clear line of sight to both?
The planet or moon may be facing the wrong way for you to be able to send a message directly to the rover or base on the surface.
Using your example, and assuming you want to send a message to a moon base on the surface of Callisto, a moon of Jupiter: while you may have a clear line to Jupiter, the point on the surface of Callisto that you want to send a message may not be in your line-of-sight because it is facing away from you. So your route becomes Earth -> Jupiter Communications Satellite 3 -> Callisto ComSat 10 -> Callisto Base Alpha
Assuming you want to send a message to the far side of Luna, even then the best path may be a commsat in high orbit around Luna -> Comsat in low orbit around Luna -> FarSide Base.
I agree, rarely are you going to need to bounce from Mars to Jupiter, but for low-priority bulk traffic (scientific images, etc) and enough nodes, it might be worthwhile to bounce from Mars to Jupiter to Luna to Earth if all communication nodes were otherwise idle. The Delay Tolerant Network Vint Cerf is working on supports knowledge of when certain antennas have available communication windows to different nodes, and can schedule and route communication accordingly.
Source: Myself, did a presentation on it in college
Assume Earth, Mars, Jupiter, and the Sun are all roughly in a line.
Let's call the amount of transmit energy needed to get a signal from Earth to Mars with it having a specified strength at the receiver 1 unit of energy.
Earth to Jupiter would require about 64 units of energy, so from the sender's point of view relaying through Mars saves them a lot of transmit energy.
The transmitter on Mars would need about 49 units of energy to get the signal to Jupiter, bringing the total to about 50 units to get from Earth to Mars.
Whether or not spending 1 unit on Earth and 49 on Mars to relay is actually cheaper than spending 64 units on Earth to go directly will depend on the cost of energy on Mars and Earth.
You don't always have a clear line of sight. E.g. when you communicate with a satellite orbiting mars, the planet will get in between for a period of time.
Furthermore, even if you have clear line of sight, repeaters on the way can be helpful because the inverse square law even holds for lasers (the light isn't perfectly parallel, there is no such solution to the maxwell equations (outside of an infinite plane emitting light perpendicular to it), instead it's a gaussian beam).
Last, you only have a limited amount of spectrum you can send to or receive from a specific point in the sky. Beyond that, it gets hard really quick. If you have repeater stations far away so that they can be targeted from the distance, you can increase the maximum bandwidth. This is mainly a concern once the interplanetary bandwidth is maxed out.
That's a routing question, nothing in the transport protocols preclude that.
If your router decides it has a direct route to Jupiter then it probably won't hop to Mars.
Well, I think that you generally don't control beam dispersion /angular resolution that well (please correct me if I'm wrong). So the received power is probably a lot better if you aim closer.
On the other hand, the latency penalty might be quite steep. There is probably a latency/bandwidth tradeoff that needs to be solved for every packet. I think today's internet has the same set of tradeoffs (plus some, like cost). If packets could carry their own metrics, routers could make better-informed choices.
We have been experimenting with DTN network since 1996 with different types of delay mechanism based on different environment settings. It’s nice to see them in action now.
I remember talking to him about this in the late 1990s. He might have been talking at Comdex or somewhere.
DTN once was a fashionable research topic in before 2012, never thought it could actually been implemented.
There was the DTNBone, but without a killer app and with the reference implementation being rather touchy and hard to debug for novices it never really went anywhere.
DTN ended up having some use cases, but they were so narrow that it never really made it out of the lab except for a handful of NASA links and some abortive military applications from people trying to solve the mobile ad-hoc routing problem.
Next they'll be saying, Bitcoin will be the interplanetary reserve asset. Then the interstellar. Intergalactic. Universal. Not in our lifetimes, is course. But it's coming.
How will you achieve consensus with years-long communication delays?
The base layer consensus protocol would run at decade speed and would just be used to net out long-term/long-distance differences; most transactions would be on higher-layer systems that settle faster. Charles Stross explains this stuff in Neptune's Brood.
How convenient that there's a fictional space opera that can verify your predictions of the future.
For what it's worth, I don't actually believe any of this. Cryptocurrency isn't the future of finance on this planet, let alone others.