Large network simulation problem

[coorasse]

Hi all!
I've got a problem simulating a network with 1000 nodes.
I've a tcl script that simulate a network with many nodes moving all around.
The trace file is of new type.
When I analyze the trace file with up to 100 nodes, the results look correct, but with 500 or 1000 nodes I can't get correct results.
For example that's are the results for traffic between node 0 and node 1 with AODV protocol:
50 nodes (1500x300):
send = 3772.00
MACpacketSend = 11645.00
recv = 3664.00
routingpkts = 8239.00
PacketDeliveryRatio = 97.14
NormalizedRoutingLoad = 2.25
RoutingOverheads = 2.18
AverageHopCounts = 3.09

100 nodes(2121x424):
send = 3796.00
MACpacketSend = 13169.00
recv = 3672.00
routingpkts = 16640.00
PacketDeliveryRatio = 96.73
NormalizedRoutingLoad = 4.53
RoutingOverheads = 4.38
AverageHopCounts = 3.47

1000nodes(6704x1342):
send = 716.00
MACpacketSend = 4759.00
recv = 610.00
routingpkts = 124574.00
PacketDeliveryRatio = 85.20
NormalizedRoutingLoad = 204.22
RoutingOverheads = 173.99
AverageHopCounts = 6.65

Obviously the last results are not correct cause node 0 can't send just 716 packets...or I'm wrong?
I attach also the awk script I use to analize results.parameters-aodv.awk

Please help me retrieving correct results, thanks.
Alessandro

[lbarreto]

Can you post here your tcl script, that way you could analyse better the
results!

Luís Barreto

On Sat, Nov 15, 2008 at 11:01 AM, coorasse <arodi@...> wrote:

>
>
> Hi all!
> I've got a problem simulating a network with 1000 nodes.
> I've a tcl script that simulate a network with many nodes moving all
> around.
> When I analyze the trace file with up to 100 nodes, the results look
> correct, but with 500 or 1000 nodes I can't get correct results.
> For example that's are the results for traffic between node 0 and node 1
> with AODV protocol:
> 50 nodes (1500x300):
> send = 3772.00
> MACpacketSend = 11645.00
> recv = 3664.00
> routingpkts = 8239.00
> PacketDeliveryRatio = 97.14
> NormalizedRoutingLoad = 2.25
> RoutingOverheads = 2.18
> AverageHopCounts = 3.09
>
> 100 nodes(2121x424):
> send = 3796.00
> MACpacketSend = 13169.00
> recv = 3672.00
> routingpkts = 16640.00
> PacketDeliveryRatio = 96.73
> NormalizedRoutingLoad = 4.53
> RoutingOverheads = 4.38
> AverageHopCounts = 3.47
>
> 1000nodes(6704x1342):
> send = 716.00
> MACpacketSend = 4759.00
> recv = 610.00
> routingpkts = 124574.00
> PacketDeliveryRatio = 85.20
> NormalizedRoutingLoad = 204.22
> RoutingOverheads = 173.99
> AverageHopCounts = 6.65
>
> Obviously the last results are not correct cause node 0 can't send just 716
> packets...or I'm wrong?
> I attach also the awk script I use to analize results.
> http://www.nabble.com/file/p20514322/parameters-aodv.awkparameters-aodv.awk
>
> Please help me retrieving correct results, thanks.
> Alessandro
> --
> View this message in context:
> http://www.nabble.com/Large-network-simulation-problem-tp20514322p20514322.html
> Sent from the ns-users mailing list archive at Nabble.com.
>
>

[coorasse ()]

That's my tcl simulation script. It's a base script without the mobility model that's it is put where you read #INITIALPOSITION.
prova1.tcl
I change the number of nodes and the topology dimensions from that script when I wanna do a different simulation.

I am thinking...it is possible that, because of the very large network topology, very few packets are sent, so the statistics are correct?

It is possible that with this sim and that awk script my calcs for the routing overhead are not correct?
However, it is possible that the number of routing packets is bigger than cbr packets number?

[coorasse ()]

up

[coorasse]

I modified a bit the file parametrs.awk but I can't post it now.
I still have problem with the scenario with 1000 nodes. Sent packets are about 30/40..and that can't be!
I also noticed that problem:
I want to add 180second at the beginning of sthe simuylation to stabilize my model, but the routing protocol does not start since I send the first cbr packet. How can I stabilize the model? Wich is the solution? I had think to start the traffic at 0sec but consider  (for analisys) only packets that have been sent after 180seconds. Can someone give me an opinion about wich can be the problem of uncorrect data results(if they are uncorrect) and about my solution to stabilze the model?
Thanks in advance,
Alessandro

[coorasse]

I tried the following:
ns-2.29-2 (precompiled from repositories)
ns-2.28
ns-2.33

and my simulation still doesn't work properly.
The results, expecially for DSR protocol are still very bad.
The ns script and the awk script are in this post.
They are not the most recent version I am using now but results doen't change.
Can someone explain me what may be the problem?
Thanks,
Alessandro

[KNOAH007]

Hi mate do u have any code for 5 nodes in wireless simulation using direct diffusion protocol

many thanx
knoah