home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: sci.crypt
- Path: sparky!uunet!haven.umd.edu!darwin.sura.net!zaphod.mps.ohio-state.edu!sol.ctr.columbia.edu!destroyer!cs.ubc.ca!uw-beaver!uw-coco!nwnexus!ken
- From: ken@halcyon.com (Ken Pizzini)
- Subject: Re: Encrypted phones, keys, taps, ...
- Message-ID: <1992Nov7.150423.26670@nwnexus.WA.COM>
- Sender: sso@nwnexus.WA.COM (System Security Officer)
- Organization: The 23:00 News and Mail Service
- References: <1992Nov6.012330.4470@shearson.com> <1992Nov7.045122.15675@netcom.com>
- Date: Sat, 7 Nov 1992 15:04:23 GMT
- Lines: 46
-
- In article <1992Nov7.045122.15675@netcom.com> rcain@netcom.com (Robert Cain) writes:
- >pmetzger@snark.shearson.com (Perry E. Metzger) writes:
- >:
- >: If "standard" means analog, well, analog phone scramblers are almost
- >: all easily broken, and are pretty poor in general. Why use one when
- >: digital techniques that are virtually unbreakable are available?
- >
- >How is this done in the case of randomly hopping variable split band
- >spectrum inverters? Current technology allows hopping between any of
- >32 frequency split points at a 60 hz hopping rate. From 16.67 ms
- >fragments how on earth does one gather any information as to its
- >content or split point?
-
- Well, first the disclaimer: Metzger did say "almost all" and "in general".
-
- > According to Micheal Washvill a voice and
- >data security specialist with a "federal agency":
- >
- > VSB scrambling can be broken but not in real time. The only
- > practical attack is through trial and error. The scrambled
- > speech must first be recorded, then divided into finite time
- > segments according to the hop rate. Each segment is processed
- > through a variety of split points until clear speech results."
- >
- >He claims a 60 to one decode time. Gimme a break! Who decides when
- >clear speech results? Automatic speech recognition is nowhere near
- >capable of this which leaves a human to determine what is clear
- >speech. Think of the combinatorics here to decide on a second! Does
- >it actually seem reasonable that this could be done in finite human
- >time or is this guy blowing self serving smoke to discourage its use?
- >
- >Or are there more sophisticated methods? I have some experience in
- >signal processing and in speech recognition and certainly can't think
- >of one. I've seen the statement that it is easy but I'd sure like to
- >see that backed up with an algorithm.
-
- If indeed "the only practical attack is through trial and error", then
- I agree that this is a pretty secure method. But what is the random
- number generator chosing the splits? If this is not cryptographically
- strong then there is a nice weakness to exploit. And keep in mind that
- to crypto folk a scheme is usually considered insecure if there is a
- way to break the code (or protocol, or whatever) that is cheaper than
- brute-force trial-and-error; "finite human time" isn't usually the metric
- used.
-
- --Ken Pizzini
-