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

Re: [EFM-Copper] Notes from 16-Oct call

Barry, Behrooz, Ken, and Vlad,

One problem today with all power-back-off testing (down and up) is
that the ITU's so-called "MAXSNRM" parameter (the maximum margin
a line should have, so that means power back off on shorter or easier
lines at the maximum-or-given data rate of the service) is clearly
specified in standards, but not implemented by anyone (or any line
code) unfortunately.

There are sometimes conflicting requirements in spectral specifications.
My experience, which has been enhanced recently by work for several
operators, is that observance of this criterion in power back-off would
greatly improve all DSL performance.  Even the latest DSL Forum
WT85, which does have a PBO test, ignores this particular parameter
of MAXSNRM.

I can't think of a better way I know, and it is simple to measure, to
assess PBO mechanisms.

John C.


At 10:55 PM 10/21/2003 -0700, Behrooz Rezvani wrote:
folks please simplify.
I suggest to have very few simple UPBO test. The functional performance of UPBO with all those tests are not really needed
 
Behrooz
 
----- Original Message -----
From: Kerpez, Kenneth
To: 'Vladimir.Oksman@infineon.com' ; barry.omahony@intel.com ; stds-802-3-efm-copper@ieee.org
Sent: Tuesday, October 21, 2003 4:40 AM
Subject: RE: [EFM-Copper] Notes from 16-Oct call

Vladimir,
 
Don't forget that we had problems measuring the received UPBO PSD in the VDSL "Olympics."
There should be a PSD mask specified at the transmitter, or at least the ability
to use narrow resolution bandwidths to measure the received UPBO PSD.
 
    Ken Kerpez
    Telcordia
    973-829-4747
-----Original Message-----
From: owner-stds-802-3-efm-copper@majordomo.ieee.org [mailto:owner-stds-802-3-efm-copper@majordomo.ieee.org] On Behalf Of Vladimir.Oksman@infineon.com
Sent: Tuesday, October 21, 2003 5:11 AM
To: barry.omahony@intel.com; stds-802-3-efm-copper@ieee.org
Subject: RE: [EFM-Copper] Notes from 16-Oct call
Dear colleagues,
 
     following the conference summary here is the proposal for VDSL UPBO assignment.
 
In accordance with the ANSI and ETSI standards, every time UPBO is turned ON, a specific UPBO mode should be defined. The mode is determined by two parameters:
1. Noise environment created by alien crosstalkers (A, F) for ANSI (998), (A, B, ... F) for ETSI (997)
2. PSD mask used (M1, M2)
 
The PSD mask is defined for all tests.
For the tests alien crosstalkers are defined (Noise A, F, ....) , the UPBO mode should be selected by the alien noise environment set.
For the tests alien crosstalkers are not defined (Self crosstalk or AWGN), use:
- for 998 and Ex mask - use UPBO for noise F
- for 998 and Cab mask - use UPBO for noise A
- for 997 and Ex mask - use UPBO for noise E
- for 997 and Cab mask - use UPBO for noise A
 
Another condition should be that the loop under test (simulation) and all the self-crosstalkers are of the same length with the same UPBO set.
 
Vladimir
 
 
 -----Original Message-----
From: O'Mahony, Barry [mailto:barry.omahony@intel.com]
Sent: Friday, October 17, 2003 5:42 PM
To: stds-802-3-efm-copper@ieee.org
Subject: [EFM-Copper] Notes from 16-Oct call
Attendees:

 

Dong Wei

Ed Eckert

Sam Heidari

Vladimir Oksman

Arthur Marris

Massimo Sorbara

Hugh Barrass

Miguel Peeters

Bernard Debbasch

Sabina Fanfoni

 

Subject: Table 62B-1 10PASS-TS Test Cases

 

The group agreed to refer to test cases by the "old" (D2.0) test numbers for the duration of the call.

 

It was noted that the test cases fall into three groups:

 

Group 1: 1-9 basic tests (10/10 Mbps)

 

Group 2:  10-20 notching off

 

Group 3:  21-31 notching on

 

In general, Groups 2 and 3 are identical conditions, except for the presence of notching.

 

It was pointed out that some of the changes made in Ancona to Group 2 were not made to Group 3.  Since these changes involved reduction in reach, this makes the table inconsistent, as the presence of notching will not increase channel capacity.

 

Accordingly, the attendees agreed that it would be desirable to change #29 to 750m, # 25 to 650m and #23 from Self noise to AWGN.  There was not unanimous opinion to delete #22, as was done with #11 in Ancona.

 

However, there was general agreement that we would focus on Groups 1 & 3, and any changes made to Group 3 would be reflected in Group 2.

 

Sabina stated we should use ETSI A/F noise models when we use ETSI profiles

 

Vladimir took the assignment to supply more detailed specification for UPBO, especially for Self and AWGN noise models, prior to the next conference call.

 

One of the unsatisfied TR's asks that simulation results or another method be presented in order to justify the numbers in the table.  In order to proceed with simulations, the assumptions must be agreed to.

 

Miguel suggested we start with Annex F of the ETSI standard [ETSI TS 101 270-1 V1.3.1 (2003-07)].  He took the assignment to come up with a proposal for assumptions for the group to use, based upon the ETSI document, prior to the next conference call.

 

Next conference call will be Thursday, 23-October, at 6 p.m. Geneva time (9 a.m. PDT).

 

 

John M. Cioffi
Hitachi America Endowed Chair Professor of Electrical Engineering
363 Packard Electrical Engineering Bldg.
350 Serra Mall
Stanford, CA 94305-9515
+1-650-723-2150  Fax: +1-650-724-3652
cioffi@stanford.edu
http://www-isl.stanford.edu/~cioffi/
http://www-isl.stanford.edu/~cioffi/dsm/