Tobias Disse, Haischel Dabian, Dan Acristinii (Kryha): Blockchain satellite comms
The team from Kryha told the Internet of Agreements conference about their rapid prototyping of blockchains for facilitating communications between satellites, based on their platform for decentralized swarm intelligence, Grex, and invited collaboration.
Transcript
Haischel: Hi guys! My name is Haischel Dabian, and he’s Dan Acristinii. We’re from a small blockchain startup based in Amsterdam, called Kryha, and on a day-to-day basis what we do is we help organisations explore blockchain, and we take it from the idea phase to actual prototypes to testing and then eventually to the pilot phase. But I’m not here to talk about our company; I’m here to talk about a project that we did at the hackathon, called Grex.ai, a decentralised hive mind.
We participated in the hackathon in April this year and where we won as well, we participated in the machine-to-machine economy track. There were different tracks, like healthcare and public safety and all of these different tracks, and we participated in machine-to-machine. What it entails is to envision a world of tomorrow and build something with the tools of today, a world where we have humans liberated from performing tedious, hard and dangerous tasks, a world where we have machines that are interacting and communicating with one another in a decentralised way.
What we came up with was Grex, grex stands for “swarm” in Latin, and what we presented was a fleet of machines that interact with one another and agree on a world state. So we had drones flying around, we had charging stations where drones can fly in, charge themselves and go execute their mission that we’re working on, we had stations that are 3D printing machines, all simulated. Basically, once you connect all of these machines with one another, you create a whole circle of their autonomous lives, where you don’t need any human interaction to keep the system running.
I’m going to show you guys a video, but just as a bit of context, what we have here is… It’s an unfortunate video, because we basically trashed all of the drones before filming this, they didn’t behave the way we wanted to. But what happened was we had drones retrofitted with Raspberry Pis where each drone is running a full blockchain node, and with this you can basically represent the world, so whatever they observe they report back to the blockchain, and in this way you can have all of the machines up to date with what’s going on. In traditional swarm intelligence you have machines that only know about their local environment, so they communicate with adjacent nodes or with what they can observe.
But what if you want to create a global intelligence? If these machines are now flying around, doing search and rescue or building or wildlife protection or planetary observation? What if you can collect all of this intelligence in a decentralised way? So, this is what we did. One drone took off, you can’t see it but there’s a marker below it, and that triggers the other two drones to basically launch and align facing the first drone. Initially we wanted to have them fly in the same direction,, but they kept crashing into each other, so we thought let’s just make them align, so that’s what the video is about.
Dan: The video you just saw is really simple: the thing is, the drones crashed so many times before we actually filmed it that they would no longer go to the same spot. The first drone you saw we already called Twitchy, because it only goes one way no matter what happens, and it’s just because of the gyroscope got completely screwed up.
We have this idea of a completely decentralised world with autonomous agents performing everything by themselves, it’s a full cycle. But to get there… We actually thought after a hackathon and everything, because the whole goal of the hackathon was what if there exists this world of tomorrow that you build with technology of today, and we’re like “Okay great, we’ve built something we can show off, but where do we go from here?” So for the following one and a half to two months we took this project and we dissected it, we’re like “What did we actually build, where are we now, and where can we go from here?” We’ve realised that great, we built that part, but that is just a world consensus, so each drone agrees on a world state. That’s great, but where do they get if from? Where is the data? How do they reach an agreement on who is allowed to access this network? Those are the other three parts.
I have a strange hobby: I really love buying domain names, especially when they’re three euros a piece, so we own all of those, I do, they’re registered to my name. [laughter] But the idea is that it actually shows you the modularity of the system. We’re not a space company; we build decentralised applications on top of existing protocols. Our devs were like “Oh yeah, it’s fine if it’s pre-alpha. If you’re okay with deploying this in 24 hours, we can build it,” and I’m like “Okay, I won’t sleep, but we did it.”
This is actually what happened, a peer-to-peer world consensus. Based on blockchain, they agree that this is the world state, and I can only add to this data, I will not be able to delete it. If you take control of a drone or if you have a malicious actor, you cannot change historical data, we don’t care. There’s another cool thing about BigchainDB for that part, is that it doesn’t use transactions but uses objects instead, because it’s kind of a database and you’re tracking an object through time. In this case it is a drone and part of a class which is the world, so you suddenly have a grid of all the drones that you always know at all times. That is the main element and why we really thought the solution was so innovative.
But before that happens we need governance, who is actually going to give you access to this data. Well, probably a DAO, probably a coop, probably a foundation. Please tell us; we know how to build it, but we’re figuring things out. Before that you need somebody… You got access to the data, but where is this data? Who guarantees you that you will be able to access this data? This is why you need a decentralised access control layer. Again, we know how to do that, but the problem is what data do you get? This is the part where we’re like “Okay, we have drones, they sense stuff, great, they can check for stuff. But if we want to do this for the entire world, the main problem is the regulations for drones, and each country is different. It’s possible, but it is so big of a headache, I don’t want to deal with this and we don’t want to deal with this, so… Oh, satellites! It’s actually an easier option if you think about it. If you do peer-to-peer data storage on satellites, where you only cache a certain level of data, you don’t need historical data. But as our potential future partner explained to us, if you have a relay and a satellite dish, you do not have to consume the entire bandwidth. So you can always keep pulling data, and you can have your own server pulling data from a decentralised system. Now you have a relay — great.
The nice part is that each of these is actually part of a protocol, and we’ve built on each of these protocols before. For decentralised data storage you need IPFS. It’s historical, it works live, it’s always there. If you have a node that pins this data, which you can actually specify, you have the data forever and you can track these changes, it’s like GitHub and BitTorrent combined. Next, for the market, it basically just means a decentralised consensus layer, which is Tendermint, it works really fast and is great and we’ve done it before. The foundation, you can do it as a DAO, but you have to be registered in Zug and there’s a lot of complications with that, we’re still figuring it out. But the point is you can only contribute data to this system if you contribute hardware, software, launches, security audits, whatever; you cannot actually buy direct access to the data.
But, if you partner with this foundation, this is the vision that we’re pushing at least, is that you get some tokens for building a small part of a satellite or launching the satellite. It might be absolutely useless for you to have a token for a data market, if we build satellite hardware. Why would you. What you do in this case is you sell yours on the secondary market. One good example is if you’re familiar with Ocean protocol, that’s how you could do it. You can always implement different ways, a secondary marketplace for data, we don’t yet know which option is best, but we know this one could work.
Afterwards you have a world state that you’ve seen before, that is BigchainDB, and the last part, what Haischel said, the final part that we really hope to build is the federated learning using blockchain, so you can actually contribute separately to a world model which is not stored in a central database nor in a distributed database which no entity has control of.
So, what happens when you take three of them? The network is based on IPFS, the market is based on Tendermint, and the foundation, which can be a DAO, it can be Ocean… Well, you buy another domain name. [laughter] Come on, it was only three euros! So, we’re like “Okay, now we can apply this to a real situation,” so we’re like “Can we do this with satellites?” We’re a blockchain company, we love building decentralised applications, but we don’t quite know how to do this. We know we can support somebody who is working on this already.
The problem is the access to space data. We really need geodata for this, but we want it in a decentralised manner, we don’t want a central exchange, because that creates a single point, and we don’t like centralised systems, it’s really not our thing. Yeah, we saw that space data is expensive, but there are alternatives. I think somebody here suggested to me the Copernicus project which also gives you free data, but again that is from a centralised part, and it’s like you’re getting the rest of the data that was not for this mission, it might not even be as up-to-date as you want, so not with the same data rate.
And yeah, space debris is a thing. This is an exaggeration, but each of these points are actually there, and it really shows the debris in low Earth orbit. And our system is really a decentralised data marketplace: why would you launch another satellite, if you can just collaborate and use part of the network? To us it seems really simple. That’s the real deal of low Earth orbit space debris — have fun with that. We really love space, we really want to be able to travel there one day, we want to be able to launch stuff up there, and if you have to calculate collisions, that’s not the best use of your time I think.
Just a couple of bullet points on why we like it. Tell us if we’re wrong, we’re really curious, but low Earth orbit is decentralised by nature. Because if you want to have an always-on system in that part, you need to have at least 50, I think the Iridium satellite network has 66, to have global coverage at all times. Access control managed on the satellites themselves: the latency involved to ask another party to get access to the network is gone, so you just get access yourself by having pre-approved sessions.
And because we use blockchain consensus algorithms, we know who is allowed to access the satellites, if the satellite you’re querying goes out of your line of sight, you can query the next one, because it already knows you’re pre-approved and it’s fine.
LEO oriented: if something goes wrong, replace it. To my knowledge, and what we figure out in the meantime — again, we’re not a space company — remote Earth sensing is performed in low Earth orbit for a lot of use cases, so a lot of people really want to use this particular orbit.
We want a data sharing platform, incentivise everybody to collaborate — it’s not an ICO, sorry! This is just an idea right now: if you really want access to the data, you have to contribute with hardware or software or whatever, and then if you’re a partner to the system, you can sell the tokens themselves, your share of access to the data. Why we really thought this would be cool is because now we suddenly don’t have to perform KYC, because the partners themselves have to perform it, so something easier for us, something we don’t want to do, something we’re not good at. We give it to participating partners that might actually be interested in performing this action themselves for getting money.
If you have questions, please ask. My closing thoughts are we’re a blockchain company, we’ve been doing this project because we’ve been tasked to do it for a hackathon and we thought it was really cool. But now, after the hackathon, we’re like “What do we do now? How do we do it? Oh, it’s applicable to space: we got invited here.” Is this relevant to you? Is anybody actually interested in going forward with this? Because we’re really interested in partnering up and pushing this further. Because we have actual building blocks available at a PoC level, but should we continue? Why? Yes? Please talk to us!
Question: Have you tried adapting this to any other systems, not just the drone systems or the satellites? The stuff you’re describing would actually fit really well for a decentralised distributed manufacturing system, where you could actually do small distributed automated factories which could do manufacturing of infrastructure.
Dan: We didn’t think about that specific use case, but we’re like “If this make sense for this situation, we could also apply it for example for autonomous cars or something, because again it’s an actual swarm. If you’re seeing elements of this that interest you, let’s talk.
Comment: I think this is really cool.
Dan: Happy to hear it!
Comment: A question from a technical point of view: where do you see yourself having a connectivity point from? Are you thinking of satellite connectivity to each of these nodes, or are you thinking more like 5G, 3G cell connectivity?
Dan: The consensus idea, before hearing the CubeSat presentation, was that there is current technology like laser satellites, and they have a peer-to-peer communication system is space. That’s why for us this was the most important layer, because once we establish consensus between satellites, then we can use relays to connect them, because we always want to have consensus up in space. So you’d probably have lasers or high frequency there, and then… You mentioned X-band I think.
Benoit: Can you work with that 100 Mbit connections?
Dan: That’s fine, 100 Mbit is great! Lower than that and you have really, really bad images, unless you’re working on really specific remote sensing data, such as temperature, pressure systems. Because that is just strings, so we can always save that to an adjacent file so we don’t need too much. When dealing with images, that is where bandwidth really matters.
Thank you very much! [applause]
Learn more http://internetofagreements.com/space
