Thread Links Date Links
Thread Prev Thread Next Thread Index Date Prev Date Next Date Index

Re: [10GMMF] FW: [10GMMF] Reminder for TP2 call, 8/26/04



Hi Jesper,

What I meant is that the analytic, or closed-form, MMSE solution (e.g.
computing the optimal tap weights based on a matrix function of the
channel impulse response) assumes a linear channel.  However, if you are
computing an adaptive (LMS) solution from the simulated waveform, then I
agree that the tap values will converge to a (potentially non-optimal)
solution.  It sounds like that is what you are doing.

Best Regards,
Paul


-----Original Message-----
From: owner-stds-802-3-10gmmf@listserv.ieee.org
[mailto:owner-stds-802-3-10gmmf@listserv.ieee.org] On Behalf Of Hanberg,
Jesper
Sent: Monday, August 30, 2004 12:35 AM
To: STDS-802-3-10GMMF@listserv.ieee.org
Subject: Re: [10GMMF] FW: [10GMMF] Reminder for TP2 call, 8/26/04

Paul

You are right about how I find the coefficients. However I do not
understand why it should not work for a nonlinear channel. The algorithm
might not find the optimal coefficients but no matter how the
coefficients are found, the simulations show that the filter is able
compensate the distorted channel.


Jesper Hanberg

Intel Copenhagen ApS
Business:       +45 44 54 61 30
Business 2:     +45 70 10 10 62
Business Fax:   +45 70 10 10 63

E-mail: jesper.hanberg@intel.com

Web Page:       http://www.intel.com


-----Original Message-----
From: owner-stds-802-3-10gmmf@listserv.ieee.org
[mailto:owner-stds-802-3-10gmmf@listserv.ieee.org] On Behalf Of Paul
Voois
Sent: 28. august 2004 09:04
To: STDS-802-3-10GMMF@listserv.ieee.org
Subject: Re: [10GMMF] FW: [10GMMF] Reminder for TP2 call, 8/26/04

Hi Jesper,

Regarding the computation of the EDC coefficients ("using MMSE"), could
you be a little more specific?  Did you use an adaptive algorithm to
compute the tap values from the simulated waveform?  Typically I think
of MMSE as referring to an analytical solution for the tap values, but
I'm not sure how that would work with a nonlinear channel like this one.

Thanks,

Paul Voois
ClariPhy

-----Original Message-----
From: owner-stds-802-3-10gmmf@listserv.ieee.org
[mailto:owner-stds-802-3-10gmmf@listserv.ieee.org] On Behalf Of Hanberg,
Jesper
Sent: Thursday, August 26, 2004 8:46 AM
To: STDS-802-3-10GMMF@listserv.ieee.org
Subject: [10GMMF] FW: [10GMMF] Reminder for TP2 call, 8/26/04

Regarding todays meeting:

EDC coefficients are obtained using MMSE  and yes see the hanberg_1_0304
presentation for BER calculation.

We are able to simulated lasers with peaking and ringing by changing the
gain compression factor. Attached is   a   figure showing   one  of the
lasers with a lower gain compression coefficient (eps_nl =2.22e-007).
Any views on how realistic this looks?

The fiber used is ranked as the 22nd worst with regards to PIE_L..

Best regards,
Jesper Hanberg
direct: +45 44 54 61 30