[sdiy] Re: Information Content of Signals

[Magnus Danielson]

From: "John L Marshall" <john.l.marshall at gte.net>
Subject: Re: [sdiy] Re: Information Content of Signals
Date: Sun, 18 May 2003 17:37:38 -0700

> But of all of the signals only one is of interest at a time. All of the
> other signals are noise to desired signal.
> 
> This is inpart how CDMA works. The desired signal is actually below the
> noise. All of the other signals are noise contributors. But, we know how to
> look for the desired signal.

Indeed. By the lovely tool of correlation impressive gains can be had.
For instance, the C/A code of L1 GPS signals, is a 1023 chip long Gold code.
This have a 30 dB gain in relation to others and the selected codes have been
chosen for best suppression of worst case sine correlation.

It is through tools like correlation you may go from a wide bandwidth to a much
more narrower while keeping the main signal intact and unwanted signal becomes
suppressed. This shift between wanted and unwanted signal is indeed a gain in
signal to noise.

TDM = Time Division Multiplexing
FDM = Frequency Division Multiplexing
CDM = Code Division Multiplexing
WDM = Wavelength Division Multiplexing

TDM, FDM and CDM is actually not as different as people is beleived to make
them. They are just three special-cases of systems having various forms of
ortogonal vectors. TDM is simple since then each have their own "timeslot".
FDM is simple since then each have their own "frequency". CDM is a little
harder yeat since then each have their own "code". They all have the property
that they have N signals where neither of them contain any part of the other,
i.e. they are linearly independent, and this also makes what happends on one of
them doesn't affect the other. Another way is to say that they do not
correlate, which is just the same. Sines of different frequency does not
correlate, and therefore is usefull as such vectors. Different sequences of
digits can be shown to have the same relative property, and thus the code they
have can be shown to have the same property. One such binary sequence is when
the n-th sequence have the n-th bit (chip) set to one and all the others set to
zero, where n is from 0 to N-1, neither of these can be expressed as the linear
sum of any other, so they do not correlate, so TDM is also such a system.

It's not all that different in theory when you toss it the right way.

WDM is just a poor choice of name since it really is FDM, the frequency doesn't
change, but wavelength might when you changes medium.

Cheers,
MAgnus