| 0:00:10 | hello i would like to talk about a later publication |
|---|
| 0:00:14 | on hacking a quantum crypto system using the to get attacked |
|---|
| 0:00:19 | we find attacking particularly interesting because one improve the systems problem is a new level |
|---|
| 0:00:25 | of security in communication |
|---|
| 0:00:28 | the data to party elephant ball once you communicate then in contrast to classical cryptography |
|---|
| 0:00:33 | one cryptography of as in the method |
|---|
| 0:00:36 | with unconditional security in principle |
|---|
| 0:00:41 | anniversary you cannot eavesdrop on the communication of elephant ball because you will introduce errors |
|---|
| 0:00:47 | and therefore |
|---|
| 0:00:48 | detected |
|---|
| 0:00:50 | as secure to prove for such a scenario requires that model assumptions what the device |
|---|
| 0:00:54 | of elephant ball |
|---|
| 0:00:57 | this works fine in theory but |
|---|
| 0:01:00 | what about in practice |
|---|
| 0:01:01 | in the next meaning will ask the question can real alice and bob communicate securely |
|---|
| 0:01:06 | and we'll show you that real device properties should be taken into account in that |
|---|
| 0:01:11 | it should prove otherwise we consider the problems |
|---|
| 0:01:17 | these are the physical clip the devices that upright and bother for task one packers |
|---|
| 0:01:21 | but obviously not possible to be shown in the publications |
|---|
| 0:01:25 | the group the system we hacked is called clarice two main console by the acoustic |
|---|
| 0:01:30 | on the left side you see the other's module on the right side you see |
|---|
| 0:01:33 | the bob model in their exchanging secret keys |
|---|
| 0:01:36 | this is how the l s module looks from the inside the bit sequence and |
|---|
| 0:01:40 | you could you need to be random so here you see to quantum random number |
|---|
| 0:01:44 | generator modules that perform this task |
|---|
| 0:01:46 | next to various optical the tectonic components you see the phase modulator that encodes the |
|---|
| 0:01:52 | bit information on optical signals |
|---|
| 0:01:54 | these optical signals are sent by alice |
|---|
| 0:01:58 | the optical fibre act as a transmission channel |
|---|
| 0:02:01 | in a practical application this is of course much longer and can be on the |
|---|
| 0:02:05 | order of fifty kilometres |
|---|
| 0:02:08 | here the optical fibre enters ball |
|---|
| 0:02:12 | well also has a face modulator to perform the basis choice |
|---|
| 0:02:17 | the two detectors in ball avalanche prototype working in gated mode |
|---|
| 0:02:22 | here we see also some additions made by us |
|---|
| 0:02:24 | so as to get some information about various electronic signals in the module |
|---|
| 0:02:29 | in the real attack the inverse three of course cannot access such information however it |
|---|
| 0:02:35 | enables us to verify that our tech really works |
|---|
| 0:02:40 | the outputs from the use electronic task and now the on the oscilloscope |
|---|
| 0:02:46 | what you see are the gate pulses applied to the two detectors |
|---|
| 0:02:50 | this tells us when detectors involve all active |
|---|
| 0:02:53 | what we also see is that alice and bob communicating frames of about thousand pulses |
|---|
| 0:03:00 | we now zoom into a single frame and see individual case |
|---|
| 0:03:04 | the system normally expect the quantum signals to arrive within these detection gates |
|---|
| 0:03:10 | our tech which is an intercept in wrist and kind of an attack times the |
|---|
| 0:03:15 | pulses however to the right after the game |
|---|
| 0:03:17 | we have checked various time delays and the intensity of the pulses to optimize the |
|---|
| 0:03:22 | attack |
|---|
| 0:03:23 | explain the principle of the attack we need some theory |
|---|
| 0:03:27 | here you can see an ivy diagram of the a p d you which we |
|---|
| 0:03:31 | can explain the guy commode of an a p d based single photon detector |
|---|
| 0:03:35 | for the gate you see that there is a bias voltage applied to the a |
|---|
| 0:03:39 | p d |
|---|
| 0:03:40 | see the look point |
|---|
| 0:03:41 | during the gate the voltage is increased about the so-called breakdown voltage see the green |
|---|
| 0:03:46 | point |
|---|
| 0:03:47 | if you foreign |
|---|
| 0:03:48 | impinge as on the a p d then an avalanche affect generates a large current |
|---|
| 0:03:53 | this generates the detection event |
|---|
| 0:03:56 | but what happens |
|---|
| 0:03:57 | if we illuminating it after the gate |
|---|
| 0:04:00 | with the bias voltage below the breakdown voltage |
|---|
| 0:04:02 | the a p d is in a linear mode |
|---|
| 0:04:05 | this means that current really a p d is proportional to the optical power |
|---|
| 0:04:10 | p d merely registers whenever the powers relatively strong |
|---|
| 0:04:14 | the detector therefore has properties |
|---|
| 0:04:16 | which are very much unlike the ones |
|---|
| 0:04:18 | during the gate |
|---|
| 0:04:19 | and also unlike the ones assumed in the security proof |
|---|
| 0:04:22 | you know application |
|---|
| 0:04:24 | we show how to exploit this deviation |
|---|
| 0:04:26 | we simulate an attack of you in which she measures all ones signal |
|---|
| 0:04:30 | and we stands bright pulses time off the box gates |
|---|
| 0:04:34 | we found that the attack when to stick one clip the system insecure for large |
|---|
| 0:04:38 | parameter range |
|---|
| 0:04:40 | so we're in contact with the manufacturer in accounting and believe that all work is |
|---|
| 0:04:45 | vital for the practical security of q k d our team currently works on the |
|---|
| 0:04:51 | other attack on that obvious to device |
|---|
| 0:04:53 | so we'll be back |
|---|