- Home Page of ProjectGuideline.com ›
- Forums ›
- Get Help to do your B.E., and M.E., M.Tech., & M.S., Project in Ns2 ›
- Bad DSR performance due to packet drop at IFQ
Hi,
I started simulating AODV, DSDV and DSR with some basic “dumb” scenarios using ns-2.29 (namely 200 nodes, high mobility, 2000×2000, 10 tcp connections at the same time, simulation time 360s) just to get a feeling for some first results.
When using DSR I realized that the amount of transmitted packets is around 8-10 times less than compared to AODV (the ratio of delivered packets “received/sent” is pretty similar). Analysis of the trace file showed that there seem to be some problems with the IFQ/MAC (802.11) as most packets are getting dropped here. Looking at the mailling-lists-archive showed that this problem seems to be known. I am not sure whether I missed some threads, but I couldn’t find a suitable solution for this problem, yet.
Can you please help and tell me which modifications I need to do to solve this problem?
best regards,
Fabian
Try to eanable the new trace format of ns2 before simulation.
While doing trace analysis, filter sent or recieved packet from suitable layer (MAC / Routing / AGT). This will avoid false counts.
Thanks for the advice, I will give it a try soon.
However, I don’t think that this will solve the problem. The problem really seems to be somewhere in the implementation of the queue. When going through the trace file everything seems to be fine up to a certain timestamp (in my case around 100s). After this timestamp there are no more data packets transmitted, but just control packets (salvation). It seems that there is a kind of buffer overflow such that there is just space to keep/store control packets in the queue which are of more importance than data packets. When going to the mailing list archive one proposal to solve this problem in DSR was to reduce the amount of packets that have to be transmitted (decreasing the transmissino rate). Unfortunately, this is not suitable for me as I want to evaluate the behaviour in scenarios of different network congestion.