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

RE: [10GBT-Modeling] RE: [10GBT-Cabling] [10GBASE-T] a channel ca pacity estim...


The cancellation levels that Solarflare used were based on measurement data.  Using these levels, in calculations based on limit lines, is meaningless.  It has been established in a tremendous series of emails that the modeling effort will use measurement data.  For those not closely watching this discussion, I refer you to the reflector archive.

To this issue of mistakenly using baud rate for analog bandwidth raises the importance of coming to consensus on the transmit PSD.  I raised this in a very early email.  I believe this will need to be addressed at the March meeting.


-----Original Message-----
From: []
Sent: Tuesday, February 25, 2003 9:46 AM
To: Fakterman, Boris
Subject: RE: [10GBT-Modeling] RE: [10GBT-Cabling] [10GBASE-T] a channel
ca pacity estim...


You are right.  The cancellation levels I used in the program is absolutely
ad-hoc.  We definitely need to agree on the reasonably achievable
cancellation levels in this group.

Also, when the baud rate is 833MHz.  The exactly capacity should be
integrated over the band [0,833/2MHz].   In the program the integration is
up to 500MHz.  This only gives a slight better result.  I believe the
difference due to the slightly larger range of integration is negligible.

Best regards,


"Fakterman, Boris" <> on
02/25/2003 03:20:37 AM

Sent by:


Subject:    RE: [10GBT-Modeling] RE: [10GBT-Cabling] [10GBASE-T] a channel
       ca   pacity estim...

Xiaopeng, and everyone

The discussed program for capacity calculations uses the cancellation
for every noise. The distributed program assumes:

Echo cancellation - 70dB
NEXT cancellation - 50dB
FEXT cancellation - 30dB
Alien NEXT cancellation - 10dB

These cancellation values have the major impact on the calculated capacity.
Probably the group should agree on the reasonably achievable cancellation
values to achieve agreement on the capacity.

Solarflare presented at January the following cancellation values:
Echo cancellation - 40dB
NEXT cancellation - 40dB
FEXT cancellation - 20dB
Alien NEXT cancellation degree was not presented.

One more comment regards the used bandwidth. While the transmitted power is
spread over 833 MHz, only 500MHz used for the analysis. The usable
for the Tx and the Rx should be 833/3 = 416.5 MHz

Fsym=833e6; %symbol rate (MHz)

Si=10*log10(10^(P/10)/Fsym)*ones(1,length(freq)); % dBm/Hz

Boris Fakterman - Intel Communications Group, Israel
Tel: 972-4-865-6470, Fax: 972-4-865-5999

-----Original Message-----
From: []
Sent: Monday, February 24, 2003 8:24 PM
Subject: Re: [10GBT-Modeling] RE: [10GBT-Cabling] [10GBASE-T] a channel
capacity estim...

Chris,  and everyone,

I have attached the latest copy of my channel capacity estimation program
to this email.  Please check it out.  If you have any program to get it,
please let me.  I can send you a copy directly.

Best Regards,

Marvell Semiconductor

(See attached file: i3e.m) on 02/24/2003 05:14:48 AM


Subject:    Re: [10GBT-Modeling] RE: [10GBT-Cabling] [10GBASE-T] a channel
       capacity estim...


Can you please reply with a copy of  the models you
had distributed in earlier e-mail?

Want to make sure a have the latest copy under discussion.



<< Albert,

 Thank you for your input.  With the consideration of ANEXT, the Shannon
 capacity of CAT-6 is 10Gbps by using the models presented in my program.
 Accordingly, your calculation shows the Shannon capacity is 12Gbps.  I
 admit this is a significant difference.  However, even with the 12Gbps
 Shannon capacity, there is no way we can achieve 10Gbps effective
 throughput over 100m CAT-6 cable if the ANEXT cannot be suppressed by any