applied cryptography - protocols, algorithms, and source code in c

Xdes2 extends this idea to a 5 round algorithm with a

Info iconThis preview shows page 1. Sign up to view the full content.

View Full Document Right Arrow Icon
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: ct to chosen-ciphertext differential cryptanalysis and found that inner-CBC is only slightly more secure than single encryption against a differential attack. If you think of triple encryption as a single larger algorithm, then inner feedbacks allow the introduction of external and known information into the inner workings of the algorithm; this facilitates cryptanalysis. The differential attacks require enormous amounts of chosen ciphertext to mount and are not very practical, but the results should be enough to give the most paranoid pause. Another analysis against meet-in-the-middle and brute-force attacks concludes that they are all equally secure [806]. There are other modes, as well. You can encrypt the entire file once in ECB, then twice in CBC; or once in CBC, once in ECB, and once in CBC; or twice in CBC and once in ECB. Biham showed that these variants are no more secure than single DES against a chosen-plaintext differential cryptanalysis attack [162]. And he doesn’t have high hopes for any other variants. If you are going to use triple encryption, use modes with outer feedback. Variants on Triple Encryption Before there were proofs that DES does not form a group, several schemes were proposed for multiple encryption. One way to guarantee that triple encryption doesn’t reduce to single encryption is to change the effective block size. One simple method is to add a bit of padding. Pad the text with a string of random bits, half a block in length, between the first and second and between the second and third encryptions (see Figure 15.2). If p is the padding function, then: C = EK3(p(EK2(p(EK1(P))))) This padding not only disrupts patterns, but also overlaps encrypted blocks like bricks. It only adds one block to the length of the message. Another technique, proposed by Carl Ellison, is to use some kind of keyless permutation function between the three encryptions. The permutation could work on large blocks—8 kilobytes or so—and would effectively give this...
View Full Document

This note was uploaded on 10/18/2010 for the course MATH CS 301 taught by Professor Aliulger during the Fall '10 term at Koç University.

Ask a homework question - tutors are online