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

[EFM] Wavelength allocation


I think that some compromise is needed on this topic. As it stands, we have
the following situation:

1) There are some standards for wavelength allocations. These standards
should be honored by every vendor.

2) As you pointed out, the actual situation is not so good as it should be,
because every vendor implement a different frequency grid.

3) As a carrier, I would like to be able to use different systems over the
same PON segment. One clever way to make it work would be by using
different wavelengths for each system. This would make the carrier's
network much more flexible and scalable, and less dependent on particular

4) Today, WDM is still in its infancy. It's expensive, and therefore more
suited to the core environment. However, with new technology, WDM may be
viable for the access network much sooner than many people expect. One
particularly good reason is the fact that the core network is inherently
limited, and all these companies doing optics need to open new
oportunities. The access front is where the volume is.

5) Of course, two laws still apply: physics and economics. It may turn out
that the development of WDM for the access is simply impossible due to a
combination of both physical and economical constraints. However, I think
that the opportunities in a flexible wavelength allocation scheme are way
too big to be ignored.

My proposal is to try to reach a compromise where we have a workable system
for the EFM needs. While I agree that G.983.3 compatibiliy would be nice,
it's not nearly as important as it may appear. I really don't imagine
mixing current DWDM traffic and EFM together over the same fiber. What I
envision is multiple EFM solutions over the same fiber, maybe co-existing
with other solutions for broadcast video. In my opinion, the frequency
allocation map for the access network may be defined independently of the
core frequency allocation map. If we manage to keep the two fully
compatible, it's much better, but it's not absolutely needed.

p.s. I know that my proposal may sound outrageous at first, but I think
that we need to open up the discussion. I'm happy to be proven wrong, given
that all the theorethical and practical reasons are well defined.

Carlos Ribeiro
CTBC Telecom