RE: Estimating the magnitude of PMD
Dave and Jonathan,
Thanks for your responses.
About integrating the PDF tail to get the 10^-12 probability, duhh...of
course. Thanks for catching it, Dave.
About using Gaussian approximation for vertical eye closure, and how to
model horizontal eye closure, I will let others comment. Collectively, it
seems like a big enough topic that we may have to kind of break it down
into modules and tackle one at a time.
About why I chose 10^-12 as my target number: I think this question will
expose a probable fallacy in my logic. I reasoned like this...at a given
wavelength, the DGD value is randomly varying in time, obeying the
Maxwell PDF. Therefore, at any instant, there is a 10^-12 probability
that this value will exceed about 5 times the mean.
Now (what I think was) the fallacy in my logic. The total energy of a
signal is divided among many wavelengths, and I assumed that all
wavelengths have the same DGD value at the same time, but of course they
don't. At every instant, every wavelength has its own probability density
function of DGD. And this integral over a two-dimensional (time,
wavelength) PDF should yield a much lower value of pulse broadening.
If I hear Jonathan, we should compute and agree on this value - soon!!
Folks, if you think you can help, please don't hesitate to contribute.
Regards,
Vipul
====================