unforgeable optical tokens?

Ben Laurie ben at algroup.co.uk
Mon Sep 23 12:54:01 EDT 2002


Nelson Minar wrote:
>>An idea from some folks at MIT apparently where a physical token
>>consisting of a bunch of spheres embedded in epoxy is used as an
>>access device by shining a laser through it.
> 
> 
> I have the pleasure of knowing one of the researchers, Ravi Pappu.
> He's smart and a real expert on holography and optics.
> 
> 
>>On the surface, this seems as silly as biometric authentication -- you
>>can simply forge what the sensor is expecting even if you can't forge
>>the token. Does anyone know any details about it?
> 
> 
> The Nature News piece claims
>   attempting to mimic the speckle pattern using some other optical
>   system, such as a hologram, is completely impractical.
>   http://www.nature.com/nsu/020916/020916-15.html
> That's obviously not a complete answer, but it suggests that the
> problem has at least been thought about.
> 
> More details are here:
>   http://web.media.mit.edu/~pappu/htm/res/resPOWF.htm
>   http://web.media.mit.edu/~pappu/htm/pubs/PappuPhDThesis01.pdf
> 
> Ravi's PhD has a section on replay attacks - section 10.3, page 135.
> The claim there is you can't store all possible challenge/response
> pairs because the keyspace is too big and that the actual system is
> too complex to simulate.

Sounds to me like you have to store a double spend database to avoid a 
replay attack (surely it isn't feasible for the verifier to choose the 
orientation with sufficient accuracy to elicit a particular response, 
therefore it will have accept valid responses from the vicinity, which 
will allow replays). And a double spend DB for this kind of thing sounds 
big and expensive. And slow to search.

Cheers,

Ben.

-- 
http://www.apache-ssl.org/ben.html       http://www.thebunker.net/

"There is no limit to what a man can do or how far he can go if he
doesn't mind who gets the credit." - Robert Woodruff


---------------------------------------------------------------------
The Cryptography Mailing List
Unsubscribe by sending "unsubscribe cryptography" to majordomo at wasabisystems.com



More information about the cryptography mailing list