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

Re: [10GBT] Pair to pair unequal SNR problem

Title: Message
Hi Eric,
  This presentation was simply to investigate how our current system performs in the presence of un-equal SNR per wire-pair ... whether from AFEXT or any other source.  If each pair were *individually* decoded, the performance would be limited to the performance of the worst case pair.  However, our current system has interleaving across pairs "built-in" to the framing.  LDPC decoding also has some amount of inherent interleaving.  With this interleaving of data from the 4 pairs, we would expect the performance of the system to be better than the SNR of the worst pair.  Our simulations confirm this but show that the overall performance is somewhat worse that what would be predicted by the average SNR.
  - Scott
-----Original Message-----
From: IEEE P802.3an Reflector [] On Behalf Of Eric Wang
Sent: Thursday, February 24, 2005 2:59 PM
Subject: [10GBT] Pair to pair unequal SNR problem

Hi Scott,
On page 10 of powell_2_0205.pdf, what you are saying is that the unequal
SNR problem is basically caused by pair to pair AFEXT differences.
Usually AFEXT is unpredictable, and the system maybe has already crashed
before it can detect which pair has the worst SNR. So I am just wondering
how interleaving and framing will help. If the unequal SNR is caused by
insertion loss differences, it is sure that simple methods, such as
interleaving, will be helpful.

Do You Yahoo!?
Tired of spam? Yahoo! Mail has the best spam protection around