QAM constellation

[M.Nasiriyan]

Hi.
Please consider two constellations of for example 16-QAM :an uniform
and a nonuniform (inwhich all of distances aren't equal) . As the
output we allocate 00, 01, 10 and 11 to the I axe respectively from
the left and the same to the Q axe from the down and we send 4
bits.This allocation is the same for both systems. So what is the
difference between two constellation for channel?

 

[Ian Stirling]

Hi.
Please consider two constellations of for example 16-QAM :an uniform
and a nonuniform (inwhich all of distances aren't equal) . As the
output we allocate 00, 01, 10 and 11 to the I axe respectively from
the left and the same to the Q axe from the down and we send 4
bits.This allocation is the same for both systems. So what is the
difference between two constellation for channel?

You want to minimise the amount of energy you send, and keep the noise
immunity the same.
The amount of noise that a uniform system can tollerate is how far from
each point you can move before crossing into another code.

If you make the constellation have a different overall shape rather than
a square, so that the individual data-bits have a more optimal shape
(closer to circular than square) then you minimise the signal power for
a given noise signal.

 

[mdp]

You begin changing the power efficiency and the individual code errors.

There are (other) motivations for utilizing a non-square QAM modulation.
For high power applications, sometimes a solid state solution is not
available. High power solutions include using a TWTA however to maximize
power efficiency one wants to run the TWTA in saturation. A square
constellation, unfortunately, wont' maintain waveform integrity passing
through a saturated amp. Consider then a 12-4 constellation consisting of 2
rings, 12 codes in the outer ring and 4 codes in the inner, can be made to
perform better through a saturated amp such as a TWTA. The outer ring is
designed to run (the TWTA) in saturation to maximize the PAE of the TWTA and
the inner ring runs (more) in the linear region of the TWTA. The codes are
not distorted and the PAE of the TWTA is maximized to the extent possible
for this application.