Boris Škorić
Assistant professor in the SEC group.
Teaching
Research
Security with noisy data
An essential property of cryptographic primitives is
an extreme sensitivity to small changes in their inputs.
However, a number of important security applications use
physical measurements as a source of (secret) randomness.
On the one hand, these measurements are inherently noisy.
On the other hand, we often wish to use them as input for
hash functions, block ciphers etc.
Some form of error correction is obviously needed
if we want reproducible results.
This requires storage/transfer of redundancy data.
It is prudent to assume that attackers have access to this data.
Hence the challenge is to develop efficient error correction
methods where the redundancy data does not compromise security.
This leads to an interesting mix of physics, information theory, coding theory and
cryptography.
The techniques developed in this field are useful in the
following applications:
- secure key storage
- anti-counterfeiting
- privacy preserving biometric identification and authentication
- true random number generation
Collusion-resistant watermarking codes
It is possible to embed hidden data in digital content such as audio and video.
This is called watermarking or fingerprinting.
In forensic watermarking a content provider embeds a unique
identification code into the content for each individual customer,
in order to be able to trace any 'leakage' of content (e.g. distribution on P2P) back to the customer.
The most powerful attack against forensic watermarks is the so-called collusion attack: multiple attackers collaborate
to remove the watermark. As they have bought differently watermarked versions of the same content, they can find the location of a significant part of their watermarks simply by comparing their content. In these locations they have a strong attack.
The content provider's defense is to use an error-correcting code
for the embedded indentifier.
I concentrate on the following questions:
- Fundamental limits on the collusion resistance of codes.
- Analysis and optimization of existing codes.
- Development of new codes.
This research topic involves information theory, statistics and analysis.
See the webpage of the
CREST project.
Computations under the encryption
Arguably the coolest thing in modern crypto:
doing computations with numbers while they are still encrypted!
MSc projects
Some ideas for (internal) master projects.
Publications
List of publications
Presentations
Presentation slides
Contact information
Boris Škorić
Security of embedded systems (SEC) group
Department of Mathematics and Computer Science
Technische Universiteit Eindhoven
tel: 040 247 4870
Visting address:
Den Dolech 2
MF 6.059
Mail address:
Postbus 513
5600 MB Eindhoven