|
![]() ![]() |
Communication Speeds, Rates and Quality A
product that we offering is based on certain mathematical considerations on the
Shannon equation, the foundation of Information Theory:
C = B log2 (1 + S/N)
(1) In
that, C is the given
channel's capacity.
(bit/s), B its bandwidth
(Hz), and S/N the
average signal-to-noise
(voltage) ratio
if the noise is additive,
“white”, and limited
in frequency.
Although Eq. (1) indicates
that doubling the channel
capacity is an impractical
task with ordinary single-ended
networks, because
of: the ensuing large
S/N reduction of about 109 dB,
it is possible
to use recently available means
with good results. More precisely,
with multitapped delay
lines as transverse filters
the ratio S/N reduces to very tolerable levels. We
have a procedure that can briefly be outlined as follows: 1.
We change the transmitted
signal symbols, or
pulses into a
series of nominally identical Gaussian
curves each alternating
in polarity. Each pulse will have a frequency
extent, let
us, say to pass 99%
of its power, double
of that of the
Channel. Thus s
=3.03/27πfc
where σ is a
pulse's half-interval between
inflection points and fc
the channel's abrupt
like in a telephone
line cutoff
frequency. The channel cannot,
of course, pass this signal
without severe and irreversible distortion
and ringing. 2.
Before transmitting the new signal
is deliberately
and reversibly pre-distorted
by halving its bandwidth with a Gaussian
filter matched to the Gaussian constituent
pulses. This operation
allows it to pass
through the given channel
without further distortion
or ringing.
3.
Finally, we recover
completely at the receiver
the original signal
by a transverse
filter, using a multitapped delay
line of specific
tap outputs, that
convolves continually
and in real time the
incoming waveform. Thereafter we continue with the normal
processes such as decoding,
etc. However, because
of the taps there is a recursive amplification
of the processed
signal and the resulting
S/N reduction is then only 30 dB. The
processes
in the above steps were reduced in a form
implementable by a computer
using software instead
of actual and common circuits.
The complete procedure is now available
in a diskette. |
Send mail to Webmaster
with questions or comments about this web site. |