Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!watmath!clyde!rutgers!lll-crg!seismo!mcvax!diku!daimi!bandholm
From: bandholm@daimi.UUCP (Anders Bandholm)
Newsgroups: sci.crypt
Subject: Re: Decryption Program
Message-ID: <423@daimi.UUCP>
Date: Tue, 28-Oct-86 08:28:07 EST
Article-I.D.: daimi.423
Posted: Tue Oct 28 08:28:07 1986
Date-Received: Wed, 5-Nov-86 06:08:35 EST
References: <2121@mtuxo.UUCP> <828@chalmers.UUCP> <3499@utcsri.UUCP>
Reply-To: bandholm@daimi.UUCP (Anders Bandholm)
Organization: DAIMI: Computer Science Department, Aarhus University, Denmark
Lines: 21

The decryption algorithm presented by Peleg & Rosenfeld in CACM Vol 22   c a n 
be implemented. 

I wrote such a program this spring. The program was written in pascal, and was
run on a DecSystem10.

Unfortunately the algorithm is very complex - both in terms of time and space.
The Dec10-version used more than 27k words of RAM. I tested the program on a 
small example in which the cipher was approx. 60 characters and where the
background text contained about 6000 characters. The background text is the
text used as "standard-english" by the program (Peleg & Rosenfeld used a novel
of about one million letters). The ciphertext was a sentence made of words taken
from the background text. After 4 iterations the text was readable. Though this
sounds as a succes, there was still one problem - the processing-time used for
the 4 iterations was about 4 hours !!!

We don't have the Dec10 any more, and I am presently translating the program 
into Modula2 for the IBM PC or perhaps for our Sun's. 

			Anders Bandholm
			(bandholm@daimi.UUCP)