Re: Why does the random input not match the output

Hi Maureen!
Great to have you here, welcome to the community!

 > There is a chunk of data missing on the bottom
 > left hand corner.

You had me scared there for a moment, but:

That's perfectly congruent with your choice of constellation points; attached output of 
(roughly this code)

import numpy
from matplotlib import pyplot
# copy & pasted from your constellation object "qam"'s constellation points:
points = numpy.array([-0.9489 -0.3162j, -0.9489 +0.3162j, -0.9489 +0.9489j, 
-0.3162-0.9489j, -0.3162-0.9489j, -0.3162-0.3162j, -0.3162+0.3162j, -0.3162+0.9489j, 
0.3162-0.9489j, 0.3162-0.3162j, 0.3162+0.3162j, 0.3162+0.9489j, 0.9489 -0.9489j, 0.9489 
-0.3162j, 0.9489 -0.3162j, 0.9489 +0.9489j])
pyplot.scatter(points.real, points.imag)
pyplot.savefig("/tmp/figure.png")

why you might not have seen the same "missing" constellation points in your channel 
model's output is the frequency offset (0.01), which rotates the whole thing by 1/100 of a 
full rotation per sample, so one "quarterrotation" every 25 samples; the synchronization 
after might be doing its best, but might not be able to counter that. I haven't actually 
looked deeper into that!

There's a few things to discuss here, like whether the constellation points are 
intentionally like they are (they might be – you might be doing an experiment where you 
intentionally confuse multiple points), and how an "unbalanced" (and hence not white) 
transmission might affect carrier recovery.

Best regards,
Marcus