biometric identification

27
Biometric identification Bozhidar Bozhanov

Upload: bozhidar-bozhanov

Post on 11-Jan-2017

6.992 views

Category:

Technology


1 download

TRANSCRIPT

Page 1: Biometric identification

Biometric identificationBozhidar Bozhanov

Page 2: Biometric identification

● @bozhobg● http://techblog.bozho.net● http://blog.bozho.net

Page 3: Biometric identification

Biometrics● Detecting inherent characteristics

○ fingerprints○ iris○ palm veins○ face○ voice○ DNA

● Unique and unchangeable

Page 4: Biometric identification

Usage● Border inspections● Access control

○ Home door unlocking● Smartphone unlocking● Looks cool in movies

Page 5: Biometric identification

Fingerprint● Binarization, thinning, extraction● Minutia (pl. minutae)

○ Ridge ending○ Ridge bifurication○ Fingerprint template

● Other methods○ Feature extraction

● MINEX (template standard)

Page 6: Biometric identification

Fingerprint

griaulebiometrics.com

binarization thinning

Page 7: Biometric identification

Storing and comparing● Original / enhanced image● Coordinates of the minutae● Other features● Fuzzy hash, locality-sensitive hash

○ “Percentage hash”○ Collisions are needed

Page 8: Biometric identification

Problems...● Bad images, dirty scanners, injured skin...

Page 9: Biometric identification

“A Japanese cryptographer has demonstrated how fingerprint recognition devices can be fooled using a combination of low cunning, cheap kitchen supplies and a digital camera.” The Register, “Gummi bears defeat fingerprint sensors”

“The results are enough to scrap the systems completely, and to send the various fingerprint biometric companies packing.” Bruce Schneier

Page 10: Biometric identification

Iris● Detection of around 200 points● Same storage methods as fingerprints● Only patented algorithms

Page 11: Biometric identification

DNA, veins, voice, face...● Using many in combination● Expensive scanners (DNA, veins)

○ But Kuwait takes DNA from everyone● Lack of uniqueness and high error rate

(voice, face)

Page 12: Biometric identification

Reconstructing● ...possible

○ based on minutae, points, features○ except if fuzzy / locality senstive hash is used

● => storing in centralized databases is dangerous

Page 13: Biometric identification

In-person verification● Easy faking+● Automated check=● Fraud

Page 14: Biometric identification

N-th factor● Secure identification is

○ something you have + ○ something you know +○ something you are

● e.g. smartcard with PIN + fingerprint (matched on the card)

Page 15: Biometric identification

Border inspections● ICAO biometric passports

○ Contain images of the face and fingerprints (soon maybe iris) (JPEG2000)

○ Integrity - with QES of the issuing authoroity● Fingerprints are read without PIN

○ ...but by a “trusted” terminal● And are compared to the person’s fingerprints● => fake/someone else’s document?

Page 16: Biometric identification

Problems● Centralized databases with images of

fingerprints● Contactless reading of fingerprints

○ 3 versions of the protocol have been demonstrated to have security issues

○ Complex scheme for certificate management. Certificates expire in 24 hours.

Page 17: Biometric identification

BSI

Page 18: Biometric identification

● ...but the chip doesn’t have a clock○ 1 leaked terminal certificate○ => all fingerprints in all passports in the world are

easy targets○ ...if the central databases don’t leak before that

● experts - “well, I can get your fingerprint from anywhere”○ in high-res?

Page 19: Biometric identification

bioID - No go● You can’t change your fingerprint/iris/DNA● Databases leak sooner or later● Easy to fake (gummi bears!)● They are used to unlock phones => unlock

○ email○ e-banking○ ...everything

Page 20: Biometric identification

Applications● 2nd factor● Border inspections with match-on-card

verification● Future?

Page 21: Biometric identification

“Free flight of the thought”● Let’s imagine...

○ Cheap and exact biometric readers● Then…

○ ID = hash(fingerprint) + hash(iris) + hash(DNA) + hash(password)

Page 22: Biometric identification

● I am 66a1aa2b4add3d8775751b81adb86e476d0a735188c2e8582be0920b2a3e55ea

● I can prove it○ scanner + app

● Distributed global electronic identity○ something I am + something I know

Page 23: Biometric identification

Fraud?● How do we guarantee that the hash is a

result of our biometrics?● biometrics+password-> KDF -> private key

(ephemeral)○ KDF (key derivation function)○ Sign challenge with the private key

Page 24: Biometric identification

Anonymity● Hashes don’t have names● Guarantees identity● Aliases for different contexts (multiple

passwords?)● Example: distributed ride-sharing with

distributed reputation system ontop of a global anonymous identity

Page 25: Biometric identification

Conclusion● Only biometrics - no● Biometrics in clear form - no● Biometrics in databases - no● 2nd factor, match-on-card - okay● Future applications

Page 26: Biometric identification

Thank you

Page 27: Biometric identification

Resourceshttp://www.theregister.co.uk/2002/05/16/gummi_bears_defeat_fingerprint_sensors/

http://www.griaulebiometrics.com/en-us/book/understanding-biometrics/types/feature-extraction/minutiae

http://www.technovelgy.com/ct/Technology-Article.asp?ArtNum=12

https://en.wikipedia.org/wiki/Key_derivation_function

http://techblog.bozho.net/electronic-machine-readable-travel-documents/

http://techblog.bozho.net/identity-in-the-digital-world/

http://europe.newsweek.com/kuwait-becomes-first-country-world-collect-dna-samples-all-citizens-and-449830?rm=eu