Friday, August 20, 2010

Lightweight cryptography for IC-printing

Yesterday I attended the first day of CHES 2010 conference. The first session brought us new results in lightweight cryptography. Quite interesting talk was given by Alex Poschmann. He proposed a new block cipher for IC-printing called PRINTcipher.

Roughly speaking IC-printing technology allows printing integrated circuits directly on a range of materials and thus the cost of production is very low. In contrast to conventional silicon manufacturing, IC-printing allows changing the printed circuits in each run with little cost. Yet another new fancy technology that should be checked in cryptographic applications especially those aimed for the RFID world.

Alex took well known strategies for designing lightweight cryptography and merge them into new design for IC-printing applications. So what is a recipe for developing a new lightweight block cipher? Lets take an SP-network and S-boxes which are as small as possible. Since we care predominantly about area, our cipher might be as slow as a turtle and iterate internal parts as long as it is needed. In the IC world, storage is costly, so we have to get rid of unnecessarily storage elements. First idea is to minimize key schedule. Yes, we can do this as long as we aim at RFIDs which are likely to have fixed keys. This means that instead of using xor gates for xoring key and data bits, we can directly use inverters and wires. Quite nice, we saved area but what
about reverse engineering and side channel attacks? This fixed key feature is most important for IC-printing. Since we are able to change and print ICs with no additional costs we can produce a bunch of RFID tags with different fixed keys very easily.

All in all, new technology, new design and slightly different approach than in traditional RFID tags. The real world applications? We will see.

No comments:

Post a Comment