Bombe Machine on an FPGA

https://people.ece.cornell.edu/land/courses/ece5760/FinalProjects/s2022/az292_kw456_lh479/az292_kw456_lh479/index.html

by vha3 • 4 years ago

83 17 4 years ago

people.ece.cornell.edu

vha3• 4 years ago

Students implemented a Bombe Machine on a DE1-SoC.

pjsg• 4 years ago

I would like to have seen the decryption of a known WWII German message to verify that all the little details are correct.

tialaramex2• 4 years ago

The Bombe doesn't decrypt Enigma encrypted messages.

What it does is figure out which keys (initial rotor positions and plugboard settings) are logically consistent with the information you have, eliminating the bulk of possibilities and leaving humans to do the remaining work. In an actual Bombe (or the replica at Bletchley today) this results in a "stop" - the machine trying every possibility halts because the combination it's trying now isn't logically impossible. Humans then take the result of the "stop" and check whether they can decrypt the message, the Bombe can continue searching for other possibilities.

The logical inconsistency results from the fact that Enigma guarantees that any symbol X will encrypt to some different symbol. It will never encrypt to X. If that seems to happen it's an inconsistency and this can't be the correct key, try again.

This is a grave problem with designs like Enigma, and it's one reason why Enigma is the dying gasp of this approach, if you've got something like Lorenz you can't rule out any of the keys by logical contradiction like the Bombes, just as you can't with modern ciphers.

nullc2• 4 years ago

To elaborate a little further:

The attack this device implements requires you to have a run of known plaintext and its corresponding ciphertext, with correct alignment between the two.

The bombe searches for keying configurations that could have produced that encipherment. When you find one, you can go test if it decrypts the rest of the message into german or gibberish.

daniel-cussen2• 4 years ago

Let's not forget one German message, one time, was undecoded to this very day. A paranoid conspiracy theorist of course, how could anyone doubt the security of the Enigma? How could anybody be onto Blechley Park's shit? A troll, if you will, in the forties.

I divine he rolled dice.

dekhn• 4 years ago

Does anybody know if something like this could be implemented using a library like Jax?

It's a bit outside my area, but I've seen a few encryption, or maybe signature breaking tools that somehow managed to apply some very cool linear algebra to guess at the bits directly using the gradient of the encryption function (can't find a nice link now)

teleforce• 4 years ago

Thanks for posting, this can be a fun and excellent entry level project for newcomers to embedded system.

Having said that, I believe that the DE1 board being used in the project is not the same as DE1 board being used in the excellent book by Prof Hamblen of Georgia Tech [1]. I'd highly recommend this book if you're starting out but the tools and the software are outdated because it is already more than a decade old.

The manufacturer of Altera embedded system boards Terasic, bizarrely decided that they wanted to re-use and retain the DE1 name although there are already DE1 and DE2 boards produced by them before this new DE1 board.

[1]Rapid Prototyping of Digital Systems: SOPC Edition:

https://link.springer.com/book/10.1007/978-0-387-72671-7

naikrovek2• 4 years ago

The board used in this project is the DE1-SoC, which is not the same as the DE1, you're right. They have different names, so I'm not sure where the confusion comes from. they definitely could have named it better, however.

djmips2• 4 years ago

I think you are sure you know where the confusion comes from. You admitted it in your last phrase?

Bombe Machine on an FPGA

Related Stories