RE: Link Model
I agree with Jonathan we need intensive experiments to learn the model, and
to confirm the model whether it is correct or needs modifications.
Our goal is to develop reliable, cost-effective products, which is no
difference from our common engineering practice, namely, the theory and test
data should always confirm each other.
In practice, there are imperfections in the manufacturing process to cause
the test results to deviate from the idealized theory. For example, the
multimode fiber is not uniformly manufactured. There are inconsistent and
defected reflective-index-profiles distributed along the longitudinal
direction of a MM fiber, which will deviate the test data from theories. In
addition, we have to deal with the DMD issues for MM fibers. Sometime, we
just can not find a model to correctly represent all the imperfections in a
Therefore, intensive test of the models and components are very important
for a reliable, cost-effective products.
It seems the contribution "300 meters on installed MMF" by Micro Linear for
November meeting is a typical example of a good mix of test data and
equations. In addition, I support 300 meter installed MMF, of which I also
have test data to support it.
Edward S. Chang
NetWorth Technologies, Inc.
[mailto:firstname.lastname@example.org]On Behalf Of Jonathan
Sent: Monday, October 25, 1999 5:38 PM
To: HSSG_reflector (E-mail)
Subject: FW: Link Model
I have had a number of questions regarding the link model published to the
reflector. From my personal records, I found a copy of the link model write
up that we had in annex A of 802.3z draft 3.2. (later deleted and not
included in the standard). This information is intended for your personal
use and not for distribution or publication.
I have deleted out the reset of clause 38 and published this annex as:
Rich Taborek did a search on related issues in the archives. These are
appended below per his append to the reflector.
Additionally, I would like to reference the following book:
"Gigabit Ethernet Networking"
Macmillan Technical Publishing;
chapter 9: "The Gigabit Ethernet Optical Link Model."
None of these sources represent a complete background on the model. Taken
together, a number of the assumptions, history, analysis, recommendations,
process, and trade-offs are still missing. Still, this is probably the best
collection of sources compiled to date for this information. There is no
tutorial available on how to use the model.
For those wanting a more detailed understanding of the model, I would highly
recommend doing what those of us who have attained such an understanding
have done: experiment with the model.
From: Rich Taborek [mailto:email@example.com]
Sent: Wednesday, September 01, 1999 3:18 AM
Subject: Re: Link Model
I spent some time researching the IEEE 802.3z presentation material and
wealth of documentation online which is available for those individuals
interested in researching the history of the GbE link model. Here's my
1) 1.25 GBAUD, 550 m, BUILDING BACKBONE LINKS ON INSTALLED 62.5 MULTIMODE
FOR IEEE 802.3:
2) Some Restricted Mode Launch (RML) Results
3) Long Wavelength Laser MMF Links: 50MMF results
4) Review of Hewlett-Packard Proposal: Long Wavelength Laser-Based Fiber
5) Chromatic Dispersion Limited Link Lengths for SWL and LWL Fiber Systems
6) Proposed Worst Case Link Model for Optical Physical Media Dependent
7) The effects of dirty connectors on laser based MMF links
8) Experimental verification of worst case optical link model
Jonathan Thatcher wrote:
> p.s. In the original note, I referenced a book that David Cunningham
> authored with William Lane. While I will not stump for the book, I know of
> no other place where either the model or the history of its development
> detailed and explained. Of what I have read, I believe the record to be
> accurate. jt
Richard Taborek Sr. 1441 Walnut Dr. Campbell, CA 95008 USA
Tel: 408-370-9233 Cell: 408-832-3957 Fax: 408-374-3645