* In a messge originally to All, Gord Hannah said: 

GH> Data compression techniques can yield additional data throughput
advantages
GH> over non-error-correcting links, by compressing data before the modem
GH> transmits it (some transfer protocols feature this ability as well).
GH> Error-correction coupled with data compression can theoretically
yield data
GH> throughputs which are many multiples of the DCE rate. It should be noted
GH> that this is accomplished by reducing the amount of data that
the modem has
GH> to transmit, via compression, not by increasing the DCE rate.
GH> 
GH> The most important question associated with a communication channel is the
GH> maximum rate at which it can transfer information. Information can only be
GH> transferred by a signal if the signal is permitted to change. Analogue
GH> signals passing through physical channels may not change arbitrarily fast.
GH> The rate at which a signal may change is determined by the bandwidth. In
GH> fact it is governed by the same Nyquist-Shannon law as governs sampling; a
GH> signal of bandwidth B may change at a maximum rate of 2B. If
each change is
GH> used to signify a bit, the maximum information rate is 2B.
GH> 
GH> The Nyquist-Shannon theorem makes no observation concerning the magnitude
GH> of the change. If changes of differing magnitude are each
associated with a
GH> separate bit, the information rate may be increased. Thus, if
each time the
GH> signal changes it can take one of N levels, the information rate is
GH> increased. As N tends to infinity, so does the information rate.
GH> 
GH> Is there a limit on the number of levels? The limit is set by the presence
GH> of noise. If we continue to subdivide the magnitude of the changes into
GH> ever decreasing intervals, we reach a point where we cannot
distinguish the
GH> individual levels because of the presence of noise. Noise therefore places
GH> a limit on the maximum rate at which we can transfer information.
GH> Obviously, what really matters is the  signal-to-noise ratio
(SNR). This is
GH> defined by the ratio of signal power to noise power and is often expressed
GH> in decibels.
GH> 
GH> There is a theoretical maximum to the rate at which information passes
GH> error free over the channel. This maximum is called the channel
capacity C.
GH> The famous  Hartley-Shannon Law states that the channel capacity
C is given
GH> by:   C  =  bandwidth  x  LOGbase2 ( 1 + SNR)
GH> 
GH> The theorem makes no statement as to  how the channel capacity
is achieved.
GH> In fact, channels only approach this limit. The task of providing high
GH> channel efficiency is the goal of coding techniques. The failure to meet
GH> perfect performance is measured by  the bit-error-rate.
GH> 
GH> THE CONNECTION PROCESS:
GH> 
GH>     Communications between computers using modems is a negotiated process.
GH> Three data transfer links are established, the DTE at the host, the DCE
GH> between the modems, and the DTE at the remote system. DTE parameters are
GH> locally established under the control of communications terminal software
GH> as limited by the capabilities of the modems. DCE parameter negotiation is
GH> somewhat more complex.
GH> 
GH>     To effect a link, several precepts must be mutually agreed to by the
GH> modems. Information regarding modulation and  error-control protocol
GH> support is exchanged between the modems, and a connection established ONLY
GH> if there is a mutually supported modulation protocol. If the modems do not
GH> incorporate a common error control protocol, the link will be established
GH> without the benefit of error control. The connect speed will be
the highest
GH> mutually supported by the modems under the common modulation protocol with
GH> the line conditions as they exist at the time of the link negotiation
GH> process.
GH> 
GH> ANSWERS TO FREQUENTLY ASKED QUESTIONS:
GH> 
GH> Question: I just replaced my trusty Generic Xpress V.32bis modem with a
GH> V.34 model, but it doesn't ever connect at 33.6Kbps. What's wrong?
GH> 
GH>    Answer: It is not only perfectly normal, but even typical in a V.34
GH>    connection to see a less than 33.6kbps connection.  V.34 is not a
GH>    fixed-speed standard, and makes/changes its connections based on phone
GH>    line quality.
GH> 
GH>    Very few people can get consistent 33.6kbps connections.  Speeds of
GH>    33.6kbps require pristine phone line quality along the entire length of
GH>    the connection.  V.34 modems are capable of pushing the
limits of analog
GH>    phone lines, commonly offering connection speeds of 21.6k, 24k, 26.4K,
GH>    28.8K, and even 31.2kbps.
GH> 
GH>    The bandwidth (or "bandpass") of a voice-grade
phone line is about 300Hz
GH>    to 3,800Hz .  Because the mathematics of encoding 33.6kbps pushes the
GH>    phone line to near its theoretical limits, V.34 was designed to
GH>    accommodate a variety of phone line conditions. V.34 is smart enough to
GH>    do what is called a "channel probe", which is a
frequency response and
GH>    signal-to-noise ratio test of frequencies at various points across the
GH>    bandpass.  During the modem handshake, the modems send a
series of tones
GH>    to each other, at known signal levels and specific frequencies.  The
GH>    modem calculates the level of the received signal at each
frequency, and
GH>    therefore can determine the maximum bandwidth available for use.
GH> 
GH>    So, just how good does a line have to be?!
GH> 
GH>    In reality, it takes line clarity at about -44dB or better (about
GH>    the sound level of a clearly whispered conversation across a
GH>    medium size room) at the top of the phone line's
"bandpass" to
GH>    obtain and maintain a 28.8kbps connection.  At about -46dB and
GH>    below, modem receivers tend to "go deaf".  The typical long
GH>    distance connection can be much worse than this at that frequency;
GH>    it is not unusual to see -55dB to -70dB (closer to the background
GH>    hiss level of a factory-fresh medium-grade audio tape).
GH> 
GH>    Standard transmit levels for domestic (US/Canada) modems are
GH> 
GH> 
GH> 
GH> 
GH>  
GH>  
---