[dccp] TFRC + real world experiments

[Ladan Gharai <ladan at ISI.EDU>]

Sally, Dccp'ers:

I have integrated Iperf with an application level implementation of  
TFRC.   This basically allows users to generate an Iperf UDP flow 
+TFRC by using a -z option. This Iperf version is available at http:// 
macc.east.isi.edu/

There are also two examples of TFRC flows generated by this code at:  
http://macc.east.isi.edu/Example1.html and http://macc.east.isi.edu/ 
Example2.html. Each example: (1) displays a single TFRC flow (with  
corresponding RTT, p, packet loss) and (2) a TFRC flow running  
simultaneously with 7 TCP flows.

The main caveat of current version of the code is that RTTs  on the  
order of 20 to 30ms cannot really be measured accurately.  Works fine  
for larger RTTs. The two shown examples have RTTs of ~77 and 122ms.

Overall the goal is to understand how a TFRC flow - per 3448 -  
behaves on the Internet, on different paths, packet sizes, cross  
flows, etc.    I hope to generate more flows on different paths, in  
particular with non-planet lab hosts. This is ongoing work and I   
would be interested in receiving comments/feedback/bug finds.

Ladan

ps: for folks unfamiliar with iperf see: http://dast.nlanr.net/ 
Projects/Iperf/