[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Meeting notes from 11/10/98



Announcements:  31 Surveys submitted by the group.  Results being tallied
for Thursdays Volunteers meeting.
There was a Security hole pointed out in SSH. Bruce is burning BSWare CD's 
with the newest kernel (2.0.36), downloaded from Red Hat.

6:30- Meeting Starts- Internet phone on Linux

Bob explained that he has not yet been successful getting a selected
internet phone package to run under Linux just yet... He's been using it 
under MS-Windows 3.11 for 3 months, and in that time it has more than 
paid for itself.. and that includes the cost of the new computer and 
accessories that go with it.  His phone bill dropped from 4 figures to under
$200.00 a month.
*and we all know it's not because he was talking any less!*  :)  

There are a lot of packages for Internet Phone out there. Bob has collected
an extensive list of URL's  for internet phone software sites, some of them
are available for Linux (or other UNIX systems). Windows and Unix 
compatibility was one of the characteristics Bob was looking for, as many
of the people he wants to talk to are Windows developers and users, 
and others are UNIX-based.

"Speak Freely" is a package that meets all of Bob's requirements, and it's
used with successfully. URL is: http://www.fourmilab.ch/speakfree
Note: This package contains encryption software, so people can speak to each
other with encrypted digital audio.

This package is done in Switzerland, where are no encryption laws.  The
current version is 6.1e, and its the first UNIX version to include an
X-Windows
interface. The MS Windows version is all one program, but the UNIX version
is several executables, with names like "sfspeaker" and "sfmike", and so on.

Since UNIX is multitasking, any of these programs can be run at the same
time, as needed.

There are also web servers where you can post your name and IP
address, so anyone can find you. One of the programs links to
such a server, also in Switzerland.

Bob explains how sound is transmitted and how a phone works...
The mechanics are the easy part.  The hard part is how to route the signal.
Since we're on the Internet, we route the sound packets the same way we
would any other TCP/IP packets -- and that's really another topic -- how 
Internet packet routing (Domain name services, IP adduces, etc.) work.

An interesting point he brought up here was that the entire Unix OS was
spawned from trying to find a solution to this problem, the telephone 
company attempted to move phone technology towards something more 
advanced than mechanical relays.

Some of the problems with moving information over the net:
It's digital, sound is analog. It's a very complex process to move sound
over the net.  Every 1/8000 of a second a chip on the sound card measures 
the voltage of the sound, taking the values and converting it to numbers,
producing a stream of bytes that can be turned into packets and sent out
over the net.

All the time these 2 sets of operations going on in your computer:
1.  Digitize your voice> Compress the sound>> Convert to Packets>>  Send it
2.  Receive the bytes>  Uncompress> Expand/convert to Analog Audio Signal.

There are two major protocols we're interested in:

TCP- Terminal Control Protocol
   Most common protocols (email or SMTP; the web, or HTTP, to name two) 
   are built in terms of TCP packets. The procedure for transmitting these 
   packets allows computers on each end to check them and ask for another
   copy of the packet if the first one didn't get there right.

UDP- User Data Protocol
   This protocol has no checking built in. The applications can do some
   checking, but many applications that use UDP don't bother, because things

   are happening too quickly. Internet Phone applications tend to use UDP, 
   since if a packet is lost, then the person listening will hear a small
gap in 
   the sound (ki d of li e try ng to re d th s!), and it's simply too late
to worry
   about the missing packets.

The question came up if people could listen in.  Yes, then can if they have
the ability to packet sniff.  There are various encryption methods that can
be used to scramble the transmission.  Among theme are:  IDEA, PGP, and DES;
Speak Freely has several encryption and compression methods.

A discussion insued then about sound cards and Linux.  Current versions of 
Linux support TRUE SoundBlaster cards (made by "Creative Labs"), although
many other brands are being supported better and better. There's also
several 
competing sound card interface and driver systems for Linux, notably "OSS"
(or
Open Sound System), and "ALSA" :for Advanced Linux Sound Architecture).