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

RE: equalization


Thanks. This is good food for thought.

Upon reflection, I think it is possible that signal distortion for
multimode fibers will be less troublesome than that created by
wireless multipath. For one thing, by restricting the excitation of
modes to a controlled subset, we may not see the occurrence of
multipath effect. The pulse may not split; it may just spread
according to the classical bandlimiting effect. In that case, we can
use spectral compensation to overcome ISI. Second, even if you see
the multipath effect, it may vary in time so slowly that we can
treat it as time invariant, unlike wireless multipath.

Of course, I agree that before we raise our hopes, we need to
scrutinize this idea.

If more than 25% of us think that there isn't enough time to reach a
judgment about this equalization idea because we need to move on
quickly, I will drop this idea. Before that, I would like us to give
the DSP folks a chance to talk to us. This is a potential solution
that can meet Objectives 1 and 2, and keep the total number of PMDs
to just 2 or 3.



> -----Original Message-----
> From: owner-stds-802-3-hssg@xxxxxxxx
> [mailto:owner-stds-802-3-hssg@xxxxxxxx]On Behalf Of
> Kardontchik.Jaime
> Sent: Monday, July 24, 2000 8:43 AM
> To: stds-802-3-hssg@xxxxxxxx
> Subject: equalization
> In a message dated 7/19/00 2:00:24 AM,
> vipul.bhatt@xxxxxxxxxxx wrote:
> >
> >Dear colleagues,
> >
> >As we think about 10G on installed MMF, there is one issue we
> >haven't discussed - equalization. Perhaps thinking about it will
> >throw more light and provide another perspective.
> >
> >In theory at least, equalization looks very promising....
> >In reality, there are a couple of challenges, applicable to both
> >850 nm and 1310 nm cases...
> >
> >1. DMD: Can equalization overcome DMD? Some have suggested
> >that DMD can be modeled as a multipath effect, something that
> >the folks in wireless industry know how to deal with. When viewed
> >in terms of a transversal filter, the multipath problem
> boils down
> >to having enough taps and setting their coefficients.
> > ...
> >I am asking if this idea is worth discussing.
> Vipul,
> It is interesting to see what the folks in the 802.11 (WLAN)
> have achieved after several years of full-time work. First,
> they (or, more correctly, their predecessors) invested a great
> amount of  years characterizing the multipath environment
> (indoor offices, buildings, factories). Then they (again their
> predecessors) developed statistical models that can reduce
> the amount of data into a small set of manageable equations
> (Rice and Rayleigh distributions) that have been found
> experimentally to represent fairly well the wireless multipath
> environment. Based on this rich amount of previous work,
> the 802.11a and 802.11b Task Forces dedicated about two
> additional years to develop Standards for the  physical layer.
> I do not see where we have this prior characterization work
> in the area of installed multimode fiber. It could take years to
> accumulate the needed data and achieve a similar degree of
> confidence in the presumable multipath-type models for MMF
> that would fit these data.
> Is there anything else we could learn from the 802.11 guys ?
> The 802.11a Task Force adopted the OFDM approach to reach
> (on paper) a respectable maximum data rate of 54 Mbps.
> In OFDM the data is sent on 48 parallel channels riding on
> different frequencies (sub-carriers). In this way the symbol rate
> on each subcarrier is smaller and a longer symbol period can
> be used. Using longer symbol periods minimizes the effects of
> multipath delay spreads. And no equalization of the multipath is
> needed (or it is a trivial one-tap equalizer after the FFT (Fast
> Fourier Transform).
> This sounds familiar. Something essentially similar is being
> proposed in 10 GbE: we call it 4-WDM. We divide the data into
> 4 parallel streams riding on four different wavelengths, so the
> symbol period in each of them is large enough that we can live
> with the spread in rise and fall times in the multimode fiber.
> The other Task Force, 802.11b took a different path: brute force
> equalization. They barely reached (on paper) an 11 Mbps
> maximum data rate. After they finished their standard
> they realized
> that they would not be able to compete with the 54 Mbps 802.11a
> Standard in the market. So they have now a new Study Group to
> enhance the speed of the 801.11b. Now, guess what ?
> In the new study group there are a lot of voices that are
> proposing
> to use OFDM (read CWDM) instead of equalization. Although there
> are additional reasons why equalization is problematic in
> high-speed
> WLAN (the need for short preambles), the general trend
> today in very
> high-speed wireless LANs is to avoid the onerous fight
> (equalization)
> against multipath and instead reduce the symbol  rate
> using parallel
> channels (OFDM) so that  the multipath effects become
> a manageable problem.
> To summarize: might be that equalization is not the right way
> to go. And even if it were - it is not a task that can be solved
> in weeks to the satisfaction of the members of the 10 GbE
> Task Force. It will not be the factor that would change
> significantly
> the results of the previous votes regarding the MMF proposals.
> On the contrary, I would expect a backfire and dissapointment
> after so many rosy expectations being raised on the Reflector.
> I would suggest to prepare the Drafts for the MMF proposals
> based on link lengths achievable without equalization. If a
> company wants to pursue the equalization path - it could do
> it and offer it as a potential differentiation advantage of its
> future 10 GbE products.
> Jaime E. Kardontchik
> Micro Linear
> San Jose, CA 95131