Slashdot Mirror
Quantum Cryptography Now Fast Enough For Video
cremeglace sends in news of a major advance in the speed of quantum key distribution. "Researchers at the Cambridge Lab of Toshiba Research Europe have solved the problem of transferring highly sensitive data at high speed across a long distance network. The team were able to demonstrate the continuous operation of quantum key distribution (QKD) — a system that allows the communicating users to detect if a third party is trying to eavesdrop on the data communication — at a speed greater than one megabit/sec over a 50 km fibre optic network, thanks to the use of a light detector for high bit rates and a feedback system which maintains the high bit rates during data transfer. ... The faster one megabit/sec data handling will allow the one-time pad to be used for the encryption of video — a vast step forward over the current ability to only encrypt voice data."
So if someone is eavesdropping, I won't be able to watch the video?
The team were...
Sorry, but that's just wrong. If they're talking about one team, then it should be "the team was" otherwise clarify it by saying "the team members were".
Want to improve your Karma? Instead of "Post Anonymously", try the "Post Humously" option.
I wonder if some interesting contributors could be noticed in founding sources...
One that hath name thou can not otter
So, do we still need the magic secondary channel which everyone doing transfers over this "theoretically perfect" channel conveniently forgets?
You only need secure transmission of keys. After that you don't care.
(I guess this is just "research"...)
No sig today...
And will remain so. Key exchange is not the issue. The issue is the symmetric encryption used afterwards (and that is present with quantum key exchange as well). Even if you disregard that, Quantum key Exchange will never be economically or security wise superior to existing solutions.
If you spend what this quantum BS costs on distributing one-time pads, you are a) provable secure b) need no new infrastructure and network links c) have no problems with routing (Quantum key exchange can only be routed optically and only for a limited distance, signal amplification is not possible) and d) spend a lot less money.
This comparison is unfair, you say, because one-time pads for n participants have size n*n? Unfortunately that is what you likely will end up for the infrastructure for Quantum Key Exchange as well, unless you have a very low number of participants. In that case the one-time pad becomes very cheap too.
Let me give you an example:
Say, we have 10 participants. Say we need 100'000 keys a day. Say a key has 256 bit, i.e. 32 bytes. A single DVD-ROM of random bits can then last for about 4 years. Generating 5GB of high-quality randomness can be done relatively cheaply, I would estimate that a generator using junction-noise can be built that gives you about 50kB/sec of random bits for less than $5000 (32 junction generators at $100 each, one 32 bit digital I/O card, one standard PC. My prototype for a junction generator is about $2 in parts, but has no shielding or filtering). That one takes a bit more than a day for the DVD. Say $10'000 overall, including labor. Then you have costs of couriering the DVDs to the destination. Say something like $100'000 per year. For a larger net, say 100 participants, use 1TB HDDs for 31 years at 1'000'000 keys/day. Or 3 years at 10'000'000 keys/day for 1000 participants.
While this is simplified, the numbers are realistic. They are several orders or magnitude cheaper than any quantum solution. Do not forget that this quantum stuff only works with people you know and that have the right (expensive) hardware already installed and are on a direct optical or optically routed link with you that is below a certain length.
And here is the killer: There are working key exchange solutions that can be made far more secure than the symmetrical encryption and that do not need any change to the network infrastructure at all. In addition, they do not have the risk that the physical theory (and it is just a theory, not fact) has a slight error that then leaks key material.
In short: This technology makes no sense whatsoever form a security or economic point of view and very likely never will.
Most ACs are not even worth the keystrokes to insult them. Be generically insulted by this and ignored otherwise.
Ridiculous.
The quantum-cryptography part is almost indubidably used for the preliminary exchange of keys.
The actual data is then sent by normal, non-quantum channels.
(1) Neither of your scenarios covers the case where both the quantum and the secondary channel are created by Eve, not just the secondary channel;
In other news, no encryption system, even some hypothetical mathematically perfect cypher, will guarantee that Bob is not actually Eve with a pair of socks stuffed down her jeans. No encryption system will tell Alice that Bob really is Bob. No encryption system will warn Alice that Bob is shagging Eve and talks in his sleep. No encryption system will warn you that Eve has tampered with your hardware. No encryption system will magically turn Alice and Bob into experienced cryptographers who will spot tampering.
Of course, you can use encryption to set up something like a trust network to validate identity, but at some point in the chain a human being has to positively identify Bob and Alice and hand them their "credentials". Likewise, no encryption system can be secure against arbitrarily sophisticated hardware/software tampering.
When you have a sexy cypher which the math says is uncrackable its easy to forget that the math depends on a whole raft of assumptions and assertions.
In a survey of 100 programmers, 111111 thought that duck-typing was a good idea.
How many attacks have been successfully executed on victims with a rock solid network? All it takes is a new bug, no patch, and a little motive... A crypto-bloated pipe would just help things because I doubt IPS technology will keep pace with this technology in the short term. Nice idea for military use, perhaps? If only secure communication is the goal it makes sense, but how is it going to solve any of today's security challenges? Feel free to prove me wrong here, I'm an Anonymous Coward after all ;)
And make sure that nobody can steal it. Isn't that the point?
Help end the use of Sigs. Tomorrow
Quantum Leap, of course. What else?
Have you driven a fnord... lately?
You must wait a little bit before using this resource; please try again later.
Then you have costs of couriering the DVDs to the destination.
Let me fix that for you :
Then you have costs of securely couriering the DVDs to the destination.
It's not a matter of just slipping the OTP DVD in a normal envelope and shipping it. You should be 100% trusting the whole route the DVD is taking, and you should be 100% trusting your storage and on-site security for the next 4 years of that DVD's useful time. This even more so as there will be a lot of DVD being transported around in your solution. You always need a secure channel, no mater what.
The trick is, with quantum key exchange, the quantum channel is inherently secure due to the laws of physic (well some attacks might still be possible depending on hardware implementation - but on the average, it's much more secure than trusting that your DVD will safely reach its destination and remain stored untampered). The other (non)-quantum channel(s) can safely be public, the quantum can't be compromised due to the way correlation work.
For average users like you this small difference is too subtle to be worth given the increases in costs.
For a small network of banks (or, in the case of another pilot project in Switzerland : for the security and privacy while routing vote results) this 100% guarantee supported by the law of quantum physics DO matter.
However slight the risk, if a OTP DVD could be compromised that's unacceptable for a small specific subset of users. And luckily, these users (banks and government) happen to be wealthy customers for quantum technology with lots of money to throw in its research. Thus even if 99.99% users out there like you and me don't give a fuck about it, the 0.01% are lucrative enough for the whole research to keep going on and for the technology to be deployed in pilot projects (between a few swiss banks and government services)
"Sufficiently advanced satire is indistinguishable from reality." - [Tips: 1DrYakQDKCQ6y52z6QbnkxHXAocMZJE61o ]
I agree with the math wizards here: It hardly matters whether this channel is secure or not since the attack will come in the form of a man-in-the-middle with both parties (incorrectly) convinced they are talking to the other. This is an attack on the certification system, not the encryption system.
With CAs already caught handing out faked certs to the authorities so they can MITM an SSL channel, the ship has already sailed on any encryption system where remote trust is required.
A brute force attack will always succeed, it will just take a long time. Never is a very long time and computers just keep getting faster.
Maybe you meant to say that there will never be a shortcut (cipher collisions, back door, etc...) to brute forcing AES128, but that is just a widely held opinion at this point, just waiting to get disproven.
Here's a quote for anybody that wants to live (and die) by their own powers of estimation:
"They couldn't hit an elephant at this dis-"
final words of General John Sedgwick, Union Commander in the U.S. Civil War, who was hit by sniper fire a few minutes after saying them http://en.wikiquote.org/wiki/Last_words
Wherever You Go, There You Are
"An underlying assumption of this analysis is that the complete keyspace is used to generate keys, something that relies on an effective random number generator. For example, a number of systems that were originally thought to be impossible to crack by brute force have nevertheless been cracked in this way because the key space to search through was found to be much smaller than originally thought, due to a lack of entropy in their pseudorandom number generators. These include Netscape's implementation of SSL (famously cracked by Ian Goldberg and David Wagner in 1995[2]) and a Debian edition of OpenSSL discovered in 2008 to be flawed.[3]"
That is to say, in the case of SSL 56 bit encryption, they used the date as a seed value and did not employ the entire 56 bits (more like 40 if I remember correctly). Deep Crack was built from custom chips built for crypto and fabbed at TSMC. The 'Unbreakable' keys ( they estimated something like the life of the Universe to crack them) were being knocked off in 56 hours.
So, I reserve the right to ignore your 'estimate' of 13 trillion years and maintain the expectation of AES128 being regularly cracked in less than a week before December of 2016
Wherever You Go, There You Are