>Otherwise, an attacker could just spoof a payment by transmitting a modified blockchain close to the payment terminal.

If the payment terminal used an SPV-like technology, the attacker would have to have a tremendous amount of computing power. This attack is not feasible, especially considering the relatively small reward of fooling an ATM or a parking meter.

Today. Parking Meters have a lifecycle of up to 25 years.

But since the meter would know the current difficulty level and with the auto-adjustment of difficulty to the amount of hash power in the network, the meter would be just as secure in 25 years as it is today even with incredible advances in hash speeds.

DVB-T tuners are pretty cheap because they're highly integrated and mass produced in huge numbers - you can get USB receiver dongles for under $10 in single quantities. There are even SoCs that integrate a processor and DVB-T demodulator on a single chip, usually used for stuff like set-top boxes.

>There are even SoCs that integrate a processor and DVB-T demodulator on a single chip, usually used for stuff like set-top boxes.

Are there any such chips other than the Broadcom chip that was announced last year?

Yeah, several companies make them, they're very popular in cheap set-top boxes. Having found what I think is the Broadcom press release you're referring to[1], the thing that's unique about their SoC is that it supports DVB-T2 which is newer and less widely used or supported. Looks like ALi have just announced a SoC for that too[2], though unlike Broadcom's it requires a seperate tuner chip.

[1] http://www.broadcom.com/press/release.php?id=s736006 [2] http://www.digitaltvnews.net/?p=24266

Oops, I read your first comment as saying "CPU, demodulator, AND tuner". Yes, there are lots of chips that integrate a CPU and demodulator, including the ultra-popular RTL2832U, very popular among SDR hobbyist (as combined with E4000 and other tuners).

The point of proof-of-work is that the attack you describe, "transmitting a modified blockchain", is very, very expensive. Essentially any plausible block is "signed & trusted" by the very-difficult partial-hash-collision embedded within.

In this situation, not really. Bitcoin's proof-of-work scheme means that if you see two different versions of the blockchain, whichever chain is longer will (with very high probability) be the globally-accepted version. But that assumes you have the ability to communicate with at least one non-malicious node.

In this case, there's a single communications channel that could be spoofed by anyone with a transmitter powerful enough to drown out the legitimate signal.

No, because proof-of-work prevents you from "faking" a blockchain. Let's say we are at block 100 and you would want to fake "the blockchain". You would have to fake a block every 10 minutes, but that requires mining, because you can't just make a block - it has to fit the previous one with the right difficulty.

You could stop sending new blocks, but that would alarm the endpoint because it will be stuck at block 100 (in our example).

The Proof-of-work scheme is pretty innovative.

It means if you could override the signal with your own AND had a lot of hash power, you could slowly drag down the difficulty of your fake chain and eventually it'd be easy to create. But indeed you must (assuming the device managed to see the real block chain at least once) do a lot of computational work to start with.

> Well, only if the transmission is signed & trusted. Otherwise, an attacker could just spoof a payment by transmitting a modified blockchain close to the payment terminal.

As noted in the article (under the "technology" page), signing is planned but not yet implemented.

> Also, things like DVB tuners are not exactly cheap enough to be integrated into many devices (though FM receivers, which may be used in the future, are).

What kind of devices are you thinking of? It seems to me that the most useful applications of this system would be for vending machines. A $10 DVB-T demodulator (at scale they can be had much cheaper than that) is a rounding error in the cost of those devices.

Not necessarily. If the installer of the machine checks that it downloaded the "real" blockchain, it would be expensive to forge new blocks to send to it.