31 October, 2005

Cryptographic Hash Workshop

A workshop on hash functions is underway today (31st of October) and tomorrow (1st of November) in Maryland, US. Sponsored by NIST, the aim is to respond to the recent collision attacks on SHA-1. Bruce Schneier has been liveblogging from the workshop.

30 October, 2005

Polish Pronounciation 101

If you've read about the history of the solution of the German Enigma machine, hopefully you'll have found out at least a little about the valuable contribution of Polish codebreakers Marian Rejewski, Jerzy Różycki and Henryk Zygalski. These mathematicians broke Enigma many years before World War II, and passed on their techniques to the British only weeks before the invasion of Poland in September 1939. This gave British codebreakers at Bletchley Park a tremendous flying-start in breaking Enigma.

That's all well and good, and is now firmly established in the literature.

But a key question remains, at least for a native English speaker: how on earth do you pronounce the names of these Polish heroes? Well, I finally grew tired of constantly butchering their names ("Marian Ray-Joo-Sky", "Jersey Rose-Icky" etc), so I asked a Polish contributor (User:Halibutt) on Wikipedia if he might record their pronounciations for me. He kindly obliged, and has so far contributed pronounciations for Marian Rejewski, Jerzy Różycki, Henryk Zygalski, Biuro Szyfrów (the Polish Cipher Bureau), Pyry (the location where the Polish passed their techniques over to the British), and Maksymilian Ciężki (head of the Cipher Bureau's German section).

To listen to the sound clips, click the link after the name on the Wikipedia article. You'll need to be able to play Ogg Vorbis files (a free audio codec), and this page gives instructions on how to do this for various common media players.

User:Halibutt might well be willing to record other Polish pronounciations if anyone has any suggestions.

29 October, 2005

Peter Gutmann's Godzilla Crypto Tutorial

Peter Gutmann's Godzilla Crypto Tutorial has been updated. This nifty resource consists of 784 slides, and whizzes through the major topics in modern, applied cryptography. It's likely to be just the thing if you need to cram crypto.

25 October, 2005

The Eight-Rotor Printing Enigma

The German military versions of the Enigma are well-known because of the historical significance of their decipherment during World War II. However, there are some other lesser-known commercial versions of Enigma, and a remarkable example of one of these is on display in a museum in Budapest (first picture, courtesy Eric Tischer). While the standard German military model had 3 rotors, and even the high-security M4 machine used on U-boat networks had 4 rotors, this rare early Enigma model had no less than 8 rotors.

A paper by Louis Kruh and Cipher Deavours ("The Commercial Enigma: Beginnings of Machine Cryptography," Cryptologia, 26(1), pp. 1–16, 2002) includes a copy of a flier for this machine, titled "The Printing Enigma". This machine, which dates from the 1920s, is distinct from two other large and bulky early commercial Enigma variants (models A and B)

According to the flyer, the machine weighed about 50kg, and measured 65cm by 45cm by 35cm (length, width and height). It printed "the plaintext in original form with letters, numbers, punctuation, word divisions". I presume it did this using a figure shift mechanism, rather than having rotors with a large number of contacts, as that the rotors seem to be labelled A-Z. The machine also printed the ciphertext into rows of 50 and groups of 5 letters. There's not much information on it cryptographically, other than the somewhat obscure claims that it had "17,576 periods / each period is 15,777,450 symbols" and that "any one of the 227,304,461,200 can be input in half a minute".

If anyone has any more information about this machine, I would love to hear about it. I'd also be interested to know the translations of the words on the large keys, which seem to be labelled "Ziffernu Zechen Zwischenraum" and "Buchstaben Zwischenraum".

22 October, 2005

KASUMI broken

KASUMI, also termed A5/3, is a block cipher used to secure 3GPP mobile phone communications. Israeli researchers Eli Biham, Orr Dunkelman and Nathan Keller have discovered a related-key rectangle attack on KASUMI that can break all 8 rounds faster than exhaustive search. The paper is to be presented at ASIACRYPT 2005 in December, but there is a technical report available online.

The attack requires 254.6 chosen plaintexts and ciphertexts, each of which has been encrypted under one of four different keys, and has a time complexity equivalent to 276.1 KASUMI encryptions. Clearly, this is not a practical attack by any stretch of the imagination (the alarmist title of this post notwithstanding), but it's an interesting result, and it invalidates some proofs about the security of the 3GPP protocols that had relied on the presumed strength of KASUMI.

Biham, Dunkelman and Keller have previously found serious flaws in the GSM stream ciphers A5/1 and A5/2.

18 October, 2005

Bletchley Park Mailing List

The Bletchley Park mailing list has thrown up some interesting bits of news in the last few days. First up was a link to a Telegraph article on the finding of the original Zimmermann Telegram decrypt. The story's a little too involved to go into here in any detail, but the gist is that during World War I, Germany got caught red-handed scheming against the United States by British codebreakers. When this was made known in the US, public opinion shifted, catalysing the entry of the United States into the war (we're told). If so, then it's quite probably the most influential single piece of decipherment in history. The news is that GCHQ have managed to unearth the original decrypt which was presented to the American ambassador in London. (A different version of the decrypt is pictured here). I was also interested to read that GCHQ have an anonymous "official historian" who's working on a secret history of the organisation.

The second bit of news is an update on the Colossus rebuild, headed by Tony Sale. According to Sale, the Psi wheels are working; previously, I gather, only the Chi wheels were operational. This means that the Colossus rebuild can now emulate the Lorenz cipher machine and decrypt ciphertext into plaintext on machine, although just one character at present! Apparently, when the Motor wheels are hooked up, the rebuild should be able to perform a standard WWII procedure whereby the Colossus was used to decrypt the first few characters of a message to check the settings on the wheels.

Thirdly and finally, it seems there's some more news at Bletchley Park. Rather than being open all year round, they are now closing for winter (at least for normal visitors), from 1st of November 2005 to 1st April 2006. Quite a shame, really. The Trust are also selling off land and leasing part of Block D to property developers English Partnerships in order to raise money. There's rumours of a change in director, too.