Hi Frank,
Lots of issues, but since I am on hold (again) with @home trying (still) to get my cable modem service to work I guess I have nothing but time! But I need to start with a health warning in response to:
You seem to have delved more deeply than the average respondent here in these matters, and I'd like to read more of your take on this topic
I am afraid that a walk through the depths of my understanding of these topics would scarcely dampen the soles of one's feet! I read and I listen (and post what I think I understand), but I don't "do", and I have tremendous respect for those of you who do. I have a strong lay-person's interest in the area, however.
IMO, these will be achieved through organic means under semiconductor-like controls more readily and more suitably (in terms of speed of switching, dynamic adaptation, etc.) than through purely optical means which employ lenses, mirrors and moving parts, hence the allure of chip based designs
I am not sure I understand this point, but if I have it right I think I agree. Filtering technology is the key to channel spacing, and if by semiconductor-type technology you mean things like the LCD-based approach being used by Chorum (and others) then we are definitely in agreement. Another perhaps as (or more) promising approach is the Mach-Zehnder (sp?) technology which Wavesplitter and others are working on. It has the advantage of being constructed from fiber, with the signal never leaving the optical path and as a result has very low insertion loss. This in turn helps reduce the power requirements.
...in the process you may have given some readers here the impression that DWDM lambdas do not employ NRZ as well. Was that your intent? In any event, please clarify that for us.
It was not my intent-in fact my understanding of existing DWDM systems is exactly the opposite. However, what I was suggesting yesterday is that a possible use of soliton modulation would be to "replace" NRZ encoding in DWDM applications, thereby enabling a better combination of speed/distance/channel count than would otherwise be possible. I may be taking an overly-simplistic view of it, but it seems as though the systems vendors (and their carrier customers) could either spend time and money on trying to remove the bad things which cause problems for NRZ-based systems at OC-192 speeds or they can go to a different encoding scheme which is less troubled by those bad things. Is there a reason that can't work?
PS-still on hold! |
