Bug 1440 - aodv throughput decrease badly from 20Mpbs to zero
aodv throughput decrease badly from 20Mpbs to zero
Product: ns-3
Classification: Unclassified
Component: aodv
Mac Intel Mac OS
: P5 normal
Assigned To: ns-bugs
: bug
Depends on:
  Show dependency treegraph
Reported: 2012-05-28 10:29 EDT by Michael Yee
Modified: 2013-04-26 14:52 EDT (History)
1 user (show)

See Also:

testing result (4.21 KB, application/vnd.ms-excel)
2012-05-28 10:29 EDT, Michael Yee
testcase (14.20 KB, text/plain)
2012-05-28 10:34 EDT, Michael Yee
output of testcase against ns-3-dev, nwifis=50 (115.40 KB, application/octet-stream)
2012-06-02 12:09 EDT, Tom Henderson

Note You need to log in before you can comment on or make changes to this bug.
Description Michael Yee 2012-05-28 10:29:05 EDT
Created attachment 1404 [details]
testing result

I use the testing code "examples/routing/manet-routing-compare.cc" in ns-3-de v and ns-3.12 to test aodv routing performance. In the simulation, the physic uses 802.11 g ERP-OFDM, set up 10 sinks, 5 Mbps for each on/off Application, Packet size is 5000 Bytes, and the simulation lasts 200 seconds, 0~50 seconds is going to setup the network topology, 51~200 seconds doing the simulation.  
     When network scale is 10 nodes, the throughput is a normal value from 51s~200s, but with 50 nodes, the performance decreases badly that the throughput decreased from 20 Mbps to 0, the throughput is close to 0 from 178s~200s and a flood of RREQ and RREP happened around 178 second region.
Comment 1 Michael Yee 2012-05-28 10:34:18 EDT
Created attachment 1405 [details]
Comment 2 Tom Henderson 2012-06-02 12:07:51 EDT
Michael, I ran your test program (after changing nWifis to 50) against the latest ns-3-dev, using the default random number seed and run number values, and did not see the performance drop that you reported.  See the output that it generated (new attachment).

Can you see whether you also obtain this result when testing against ns-3-dev?  I did not test against ns-3.12, because that version of aodv did have some bugs (since fixed) that might explain what you observed.
Comment 3 Tom Henderson 2012-06-02 12:09:11 EDT
Created attachment 1410 [details]
output of testcase against ns-3-dev, nwifis=50