maximizing cryptographic return

[John A. Limpert]

At 06:49 PM 9/3/95 -0700, Vladimir Z. Nuri wrote:
>the question of the cost-effectiveness of  phone encryption was raised by 
>my other message. I would like to question how cheaply good encryption
>could be done on phones, with a poor quality microprocessor.

It doesn't take much of a microprocessor to do real-time encryption and
decryption of digitized voice, at least for DES. The problem is the
vocoder. A good vocoder needs a fast DSP chip. A custom vocoder chip
could cut this cost considerably, look at how cheap data pumps for
V.34 modems have become with mass production. If ISDN ever becomes the
mass market standard, replacing analog local loops, it would be very
inexpensive to add encryption.

Even with today's phone system, I can't see why a mass market secure
telephone couldn't be built for less than $200. The problem is
convincing enough people that they need secure telephones to get
that mass market.

The secure telephone also needs to be transparent to the users.
It should be able to setup a secure connection without requiring
the user to press any buttons or know what hardware is at the other
end. The voice quality has to be much better than some of the
"Donald Duck" quality systems that have been used in the past.


--
John A. Limpert
johnl at Radix.Net