Thanks for the reply, i've just found the problem, i haven't included the "Flags header" and no error occurs but the results was totally wrong
then my module to use ecn have to set the ECN bit in the IP header (can i do the same things that are done by the RED queue management?) kind regards Davide 2008/12/18 Sally Floyd <sallyfl...@mac.com>: > Davide - > >> >> ---------- Forwarded message ---------- >> From: Davide <jimi...@gmail.com> >> Date: 2008/12/17 >> Subject: ECN enabling >> To: ns-users <ns-users@isi.edu> >> >> >> hi, >> >> i'm using tcp reno and i want to turn on the ecn_ feature. i'm using >> all module that don't use this information (the queue are all drop >> tail) and only one module can set this flag. >> >> to enable ecn i used this function >> >> Agent/TCP set ecn_ 1 >> >> and even if my module don't send any ecn notification the result of >> the simulation are totally different from the case with disabled ecn >> >> there is something else that i have to know? >> i've also took a look inside the test suite but in that case the ecn >> notification are generated from the RED queue management >> >> any help appreciated > > One thing to do would be to look at the trace file, and > see if any packets have the Congestion Experienced bit > set in the ECN field of the IP header. > > From the manual, the flags in the trace file are as follows: > > "The next field contains the flags, which not used in this example. > The flags are defined in the flags[] array in trace.cc. Four of the > flags are used for ECN: ``E'' for Congestion Experienced (CE) and > ``N'' for ECN-Capable-Transport (ECT) indications in the IP header, > and ``C'' for ECN-Echo and ``A'' for Congestion Window Reduced (CWR) > in the TCP header. For the other flags, ``P'' is for priority, and > ``F'' is for TCP Fast Start." > > If the Congestion Experienced bit is not being set, the next thing > would be to see if TCP is setting the ECN-Echo bit. It is possible > that when TCP is ECN-capable, the TCP implementation in ns2 uses > ECN-Echo even when Congestion Experienced is never set by any routers > along the path - I forget exactly. If so, it could affect TCP > performance. > > It is also the case that different runs of a simulation can > get different results because they use different seeds > for the random number generator... > > - Sally > http://www.icir.org/floyd/ > >
