4 ms·

A blockchain, needing to be received by every user at the same time, is literally the perfect use case for satellite broadcasting. It was just a matter of time

by mrb 9y ago

A blockchain, needing to be received by every user at the same time, is literally the perfect use case for satellite broadcasting. It was just a matter of time that someone was going to do it. Jeff Garzik has (had?) plans to launch micro Bitcoin satellites... And, no, Blockstream didn't launch their satellites, they simply used an existing satellite service provider. I used to work for Smartjog who did the same. In fact I wrote the software (RBC, Reliable Bit Cast) used to send the actual satellite data packets through our provider Intelsat. The way it works is pretty simple. We had a server collocated in Intelsat's "teleport" in Riverside (http://www.intelsat.com/global-network/intelsatone/teleports/ http://www.intelsat.com/global-network/intelsatone/teleports...). I walked into their data center, racked our server, hooked it up (yeah I did IT too, we were a small company back then.) And they tell you to send a stream of UDP/IP packets to one of their TCP endpoints, which ingests data into their satellite infrastructure. On the receiver's end all you need is a standard dish hooked to a DVB card. Can't remember which one we used, but it had a Linux driver that presented a virtual network interface, and Linux received the IP traffic just as if it was a standard Ethernet NIC. I don't know the cost of this Intelsat service, but it can't be that high, maybe $5-20k/month for the lowest bandwidth option. On another side note, more companies broadcasting the blockchain and covering more of the planet (the real time map shows Asia as not covered) will eventually put the "block propagation delay" argument in the tomb, for those arguing against big blocks.

nosuchthing 9y ago
I can't imagine any benefits from satellites in blockchains unless everyone has access to broadcast their transactions which would seem to necessitate satellite uplink hardware for any node which I highly doubt will become affordable or common anytime soon and until it does, the nodes running sattalite uplinks would be in a position similar to mining pool operators which can MITM your transactions to manipulate for their malicious benefit. Also latency on satellites is atrocious.
- mrb 9y ago
  The sat link is downstream only. The benefit is obvious: you can run a full node with the cheapest lowest-bandwidth internet connection, while shifting the majority of your network traffic to the (free) sat link. You use your regular internet connection only for the occasional retransmissions, or transactions you emit, or blocks you mine.
  - icebraining 9y ago
    And you don't even need an internet connection to push your transactions: http://gk2.sk/how-to-push-bitcoin-transactions-via-sms/ http://gk2.sk/how-to-push-bitcoin-transactions-via-sms/ (Not my blog, I just had the same idea on reading the article)
- nadaviv 9y ago
  Broadcasting transactions is easy and trustless: you can give it to anyone who has an internet connection (over radio, SMS, in-person, physical mail, pigeons...) and the worse they can do is not broadcast the transaction. If you send the transaction to multiple parties, you can reduce this risk without introducing any additional risk. It's also something that you don't do very often. Figuring out Bitcoin's best chain (PoW-wise longest) and remaining in consensus with the rest of the network is the hard part: it requires on-going communications on a reliable uncapped/high-capped internet connection, introduces trust, and adds the risk of fraud (someone feeding you with an alternative best chain that's not recognized by the rest of the network as such could execute double-spend attacks against you [0]). [0] Blockstream satellite is no different in this regard: Blockstream could feed incorrect data into their satellites. But as a farmer in India, I would feel more comfortable trusting a satellite signal from an internationally known company that many other people are also getting (and audit, and hopefully sound an alarm if something goes wrong) then trusting, say, some internet-connected device two villages away.
  - nadaviv 9y ago
    Also, one could use the satellite data as an additional source of data, just as-if its another peer on the p2p network. In this case, this would strictly improve your security and protect against partitioning attacks with no downsides at all.
repomies691 9y ago
> A blockchain, needing to be received by every user at the same time, is literally the perfect use case for satellite broadcasting. I have never seen any interest from governments actually blocking blockchain data, or in general censoring bitcoin traffic. Using bitcoin via internet will highly probably stay the cheapest and easiest method in the future, unless really there is need to have millions/billions of nodes. Doesn't look to me like that, in fact I think quite many bitcoin nodes are just hobbyist nodes that don't actually do any payment processing except for few random personal transactions. So it is quite difficult to see the applications for this, unless made "just for fun". These projects are just cool PR stunts. And they actually seem to work as such pretty well.
- kobeya 9y ago
  Is your internet connection free?
xorcist 9y ago
Not sure if the propagation delay is sarcastic or not (hard in writing), but network latency is worse with satellite links. Which then doesn't matter because network propagation is largely set by verification worst cases.
- mrb 9y ago
  I was not sarcastic. Studies like http://www.tik.ee.ethz.ch/file/49318d3f56c1d525aabf7fda78b23fc0/P2P2013_041.pdf http://www.tik.ee.ethz.ch/file/49318d3f56c1d525aabf7fda78b23... have shown that 5% of nodes with a regular internet connection still haven't received the block after 40 seconds; and the median time is around 13 seconds. The sat latency at 1-2 seconds is much lower and would hugely improve block propagation delays. Edit: a second paper (2016) has similar findings: 10% of nodes haven't received the block after 79 seconds: http://fc16.ifca.ai/bitcoin/papers/CDE+16.pdf http://fc16.ifca.ai/bitcoin/papers/CDE+16.pdf
  - tlrobinson 9y ago
    This paper was written in 2013, before any relay network was in place (originally the Bitcoin Relay Network http://bitcoinrelaynetwork.org/ http://bitcoinrelaynetwork.org/ currently FIBRE http://bitcoinfibre.org/ http://bitcoinfibre.org/ and others)
phillc73 9y ago
> I used to work for Smartjog who did the same. In fact I wrote the software (RBC, Reliable Bit Cast) used to send the actual satellite data packets through our provider Intelsat. I really loved Smartjog for a while and even commissioned a couple of devices in the UK. It was a real shame the likes of Aspera and Signiant won that space. I seem to recall the RBC library was open source. Is that still available somewhere?
- mrb 9y ago
  Yes RBC was eventually opened sourced, long after I quit Smartjog in 2004, but I don't know if it is still available somewhere.
kobeya 9y ago
> On another side note, more companies broadcasting the blockchain and covering more of the planet (the real time map shows Asia as not covered) will eventually put the "block propagation delay" argument in the tomb, for those arguing against big blocks. Um, no. First of all the limiter needs to be the decentralized internet p2p relay, since even a federation of satellite broadcasting companies is unacceptable. This is adversarial engineering where you design to operate in the worst case. Second, it is validation of a block that matters, not mere receipt of the header. So how long it takes to transmit a block over the low bandwidth connection such a satellite provides, plus the time to validate CAN be significant as you increase the blocksize beyond the current 4MB.
- mrb 9y ago
  "First of all the limiter needs to be the decentralized internet p2p relay," The p2p overlay network should remain the backbone of Bitcoin. It is perfectly adequate for most nodes, however critics of large blocks have used the argument that a small percentage of nodes just don't even have adequate internet bandwidth. Now a satellite service gives them the guarantee they can receive blocks, no matter their size (bandwidth is sized appropriately), very cheaply, no matter where they are located in the world, no matter how bad their internet connection is. "Second, it is validation of a block that matters" I never said this satellite service will magically fix everything, just that it removes ONE hurdle of many.

by mrb 9y ago

nosuchthing 9y ago

I can't imagine any benefits from satellites in blockchains unless everyone has access to broadcast their transactions which would seem to necessitate satellite uplink hardware for any node which I highly doubt will become affordable or common anytime soon and until it does, the nodes running sattalite uplinks would be in a position similar to mining pool operators which can MITM your transactions to manipulate for their malicious benefit. Also latency on satellites is atrocious.

mrb 9y ago
The sat link is downstream only. The benefit is obvious: you can run a full node with the cheapest lowest-bandwidth internet connection, while shifting the majority of your network traffic to the (free) sat link. You use your regular internet connection only for the occasional retransmissions, or transactions you emit, or blocks you mine.
- icebraining 9y ago
  And you don't even need an internet connection to push your transactions: http://gk2.sk/how-to-push-bitcoin-transactions-via-sms/ http://gk2.sk/how-to-push-bitcoin-transactions-via-sms/ (Not my blog, I just had the same idea on reading the article)
nadaviv 9y ago
Broadcasting transactions is easy and trustless: you can give it to anyone who has an internet connection (over radio, SMS, in-person, physical mail, pigeons...) and the worse they can do is not broadcast the transaction. If you send the transaction to multiple parties, you can reduce this risk without introducing any additional risk. It's also something that you don't do very often. Figuring out Bitcoin's best chain (PoW-wise longest) and remaining in consensus with the rest of the network is the hard part: it requires on-going communications on a reliable uncapped/high-capped internet connection, introduces trust, and adds the risk of fraud (someone feeding you with an alternative best chain that's not recognized by the rest of the network as such could execute double-spend attacks against you [0]). [0] Blockstream satellite is no different in this regard: Blockstream could feed incorrect data into their satellites. But as a farmer in India, I would feel more comfortable trusting a satellite signal from an internationally known company that many other people are also getting (and audit, and hopefully sound an alarm if something goes wrong) then trusting, say, some internet-connected device two villages away.
- nadaviv 9y ago
  Also, one could use the satellite data as an additional source of data, just as-if its another peer on the p2p network. In this case, this would strictly improve your security and protect against partitioning attacks with no downsides at all.

repomies691 9y ago

> A blockchain, needing to be received by every user at the same time, is literally the perfect use case for satellite broadcasting. I have never seen any interest from governments actually blocking blockchain data, or in general censoring bitcoin traffic. Using bitcoin via internet will highly probably stay the cheapest and easiest method in the future, unless really there is need to have millions/billions of nodes. Doesn't look to me like that, in fact I think quite many bitcoin nodes are just hobbyist nodes that don't actually do any payment processing except for few random personal transactions. So it is quite difficult to see the applications for this, unless made "just for fun". These projects are just cool PR stunts. And they actually seem to work as such pretty well.

kobeya 9y ago
Is your internet connection free?

xorcist 9y ago

Not sure if the propagation delay is sarcastic or not (hard in writing), but network latency is worse with satellite links. Which then doesn't matter because network propagation is largely set by verification worst cases.

mrb 9y ago
I was not sarcastic. Studies like http://www.tik.ee.ethz.ch/file/49318d3f56c1d525aabf7fda78b23fc0/P2P2013_041.pdf http://www.tik.ee.ethz.ch/file/49318d3f56c1d525aabf7fda78b23... have shown that 5% of nodes with a regular internet connection still haven't received the block after 40 seconds; and the median time is around 13 seconds. The sat latency at 1-2 seconds is much lower and would hugely improve block propagation delays. Edit: a second paper (2016) has similar findings: 10% of nodes haven't received the block after 79 seconds: http://fc16.ifca.ai/bitcoin/papers/CDE+16.pdf http://fc16.ifca.ai/bitcoin/papers/CDE+16.pdf
- tlrobinson 9y ago
  This paper was written in 2013, before any relay network was in place (originally the Bitcoin Relay Network http://bitcoinrelaynetwork.org/ http://bitcoinrelaynetwork.org/ currently FIBRE http://bitcoinfibre.org/ http://bitcoinfibre.org/ and others)

phillc73 9y ago

> I used to work for Smartjog who did the same. In fact I wrote the software (RBC, Reliable Bit Cast) used to send the actual satellite data packets through our provider Intelsat. I really loved Smartjog for a while and even commissioned a couple of devices in the UK. It was a real shame the likes of Aspera and Signiant won that space. I seem to recall the RBC library was open source. Is that still available somewhere?

mrb 9y ago
Yes RBC was eventually opened sourced, long after I quit Smartjog in 2004, but I don't know if it is still available somewhere.

kobeya 9y ago

> On another side note, more companies broadcasting the blockchain and covering more of the planet (the real time map shows Asia as not covered) will eventually put the "block propagation delay" argument in the tomb, for those arguing against big blocks. Um, no. First of all the limiter needs to be the decentralized internet p2p relay, since even a federation of satellite broadcasting companies is unacceptable. This is adversarial engineering where you design to operate in the worst case. Second, it is validation of a block that matters, not mere receipt of the header. So how long it takes to transmit a block over the low bandwidth connection such a satellite provides, plus the time to validate CAN be significant as you increase the blocksize beyond the current 4MB.

mrb 9y ago
"First of all the limiter needs to be the decentralized internet p2p relay," The p2p overlay network should remain the backbone of Bitcoin. It is perfectly adequate for most nodes, however critics of large blocks have used the argument that a small percentage of nodes just don't even have adequate internet bandwidth. Now a satellite service gives them the guarantee they can receive blocks, no matter their size (bandwidth is sized appropriately), very cheaply, no matter where they are located in the world, no matter how bad their internet connection is. "Second, it is validation of a block that matters" I never said this satellite service will magically fix everything, just that it removes ONE hurdle of many.