Jon,
For Web traffic, a practical goal would be that almost all (say
99%) of Web pages load without any bit errors. Considering the
RTT, a single error in any packet in any object in the page would
probably make that page finish loading noticeably slower. 1E-8 is a
good rough rule of thumb for this objective.
Of course, achieving 1E-8 with only convolutional encoding can get
expensive in terms of antenna size or transponder power, so
Reed-Solomon FEC almost always makes sense. Once you decide to use
R/S, the rolloff of BER with reduced Eb/No is so steep that you can
easily get 1E-8 as long as the link is up at all, and even better
BER for little extra power.
Regards,
Ralph Brooker
Andrew Corp.
______________________________ Reply Separator
_________________________________
Subject: BER and TCP/IP performance
Author: "Jon Mansey" [SMTP:[email protected]] at AOP
Date: 3/30/99 4:09 PM
Hi,
Can anyone point me to some research or tables showing how BER affects
TCP/IP thoughput?
At the same time, does anyone have any real world experience of what is an
acceptable BER to shoot for in designing end-end sat links for TCP/IP.
Thanks in advance.
Jon.
[email protected] Chief Science Officer
------------------------------------------------------------------
"Low-cost & high-speed access to the US Internet via Satellite"
InterPacket Group, Inc. http://www.interpacket.net
1901 Main St. 2nd Floor tel (310) 382 3300
Santa Monica, California 90405 fax (310) 382 3310
------------------------------------------------------------------
"Unix IS user friendly... It's just selective about who its friends are."
This archive was generated by hypermail 2b29 : Mon Feb 14 2000 - 16:14:54 EST