From: jthomas@kolanut.mitre.org (Joe Thomas)
To: cypherpunks@toad.com
Message Hash: 693fd98f5462cb1e27cecbe955e67262c1cd7801e7e445ebabaa095fa738e904
Message ID: <9308121503.AA00397@kolanut>
Reply To: N/A
UTC Datetime: 1993-08-12 15:08:12 UTC
Raw Date: Thu, 12 Aug 93 08:08:12 PDT
From: jthomas@kolanut.mitre.org (Joe Thomas)
Date: Thu, 12 Aug 93 08:08:12 PDT
To: cypherpunks@toad.com
Subject: Re: Secure voice software issues
Message-ID: <9308121503.AA00397@kolanut>
MIME-Version: 1.0
Content-Type: text/plain
Phil Karn <karn@qualcomm.com> wrote:
[Re: squeezing 13.3 kbps data w/o start&stop bits over modem]
> Whenever V.42 error correction (LAPM) is enabled,
> synchronous HDLC frames are what actually pass over the
> link. So the start/stop bits are already removed.
> Unfortunately, the packetizing done by LAPM adds delay
> we don't want for a real time voice application. And if you
> turn off LAPM, you return to sending the start/stop bits
> over the wire.
Well... How much latency does LAPM really add? Little enough that
full duplex keystrokes echo back nearly instantly on my connections.
People talk over satellite links with more delay than that all the
time. Since we're not going to get toll quality voice out of the
vocoders anyway, and the error correcting stuff is especially useful
for encrypted data streams... I think V.42 is probably a good idea
for a cryptophone project.
> Most V.32 and V.32bis modems provide for direct
> synchronous operation, which would let us have our cake
> and eat it too, except that few PCs can speak
> synchronously to a modem. This may require some extra
> hardware (sigh).
I'm in favor of getting a minimal version that will run on the lowest
common denominator hardware first. (Might have something to do with
the fact that I just spent a couple hundred dollars on an internal
V.32bis modem that doesn't do synchronous :^)
Joe
Return to August 1993
Return to “karn@qualcomm.com (Phil Karn)”