[e2e] Outliers of delays on the last (wireless) hop and the I nternet

[Tom Mchugh]

Hi Vasilis,
I dont know if these would be interesting to you (I dont remember a whole
lotta "pdf" but some of the other stuff is there):
"End-to-end Reliability in UMTS: TCP over ARQ, A-F.Canton and T.Chahed,
Institute National des Telecomminications"
and Gorry&Mahesh's "Implications of Broadband Satellite BOD Algorithms on
TCP Performance" (sat stuff, what do you expect - sorry:)
tom_mchugh at inmarsat.com <mailto:tom_mchugh at inmarsat.com> 

-----Original Message-----
From: end2end-interest-admin at postel.org
[mailto:end2end-interest-admin at postel.org]On Behalf Of Vasilis Friderikos
Sent: Tuesday, March 05, 2002 12:26
To: end2end-interest at postel.org
Subject: [e2e] Outliers of delays on the last (wireless) hop and the
Internet
Importance: High



I know that the topic that I will touch in the next few lines is more close
related to the PILC list, but the end2end list (traditionally) has a more
wider scope. There is a lot of research effort to characterize the TCP
performance in the last wireless hop ( a lot of solutions[which some of them
break the end-to-end principle]) but it seems that there is a lack of more
realistic performance evaluation where both the Internet and wireless
dynamics are taken into account. Mobile hosts which are connected to a Web
server which is let's say 20-25 hops away will experience both kind of
dynamics. So, it seems to me that we can characterize the profile of this
delay by a convolution of the delays in the Internet and those due to the
fluctuations on the wireless link (even though there is no such global pdf
that express the delays on the Internet,[i.e NLANR active measurements].But,
I'm not aware of  pdf's of the delays on the last hop for different access
schemes( pdf that take into account the specific MAC used,ARQ,
transmission,coding-decoding).Also,concerning TCP issues, a knowledge of
possible outliers, the frequency of these outliers together with the
dynamics of the delays on the Internet would give a better (I think)
understanding of TCP{flavours} performance. If there is a work that combines
these two "worlds" I would like to be aware of. Thanks in advance.
 
Vasilis
 
 



**********************************************************************
This email and any files transmitted with it are confidential and
intended solely for the use of the individual or entity to whom they
are addressed. If you have received this email in error please notify
the system manager.

In accordance with Inmarsat Information Security Policy and Guidelines 
on Computer use, emails sent or received may be monitored. 

Inmarsat Limited 99 City Road London EC1Y 1AX.
Registered in England and Wales No. 3675885
**********************************************************************

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://www.postel.org/pipermail/end2end-interest/attachments/20020305/2a49bb35/attachment.html