aomdv problem

[eyasu tadese]

root@fe6cae3aeef0:/home/ns3_project/ns-allinone-3.26/ns-3.26# ./waf --run aomdv-example
Waf: Entering directory `/home/ns3_project/ns-allinone-3.26/ns-3.26/build'
Waf: Leaving directory `/home/ns3_project/ns-allinone-3.26/ns-3.26/build'
Build commands will be stored in build/compile_commands.json
'build' finished successfully (45.983s)
Creating 10 nodes 100 m apart.
Starting simulation for 10 s ...
PING  10.0.0.10 56(84) bytes of data.
assert failed. cond="cur->tid != tag.GetInstanceTypeId ()", msg="Error: cannot add the same kind of tag twice.", file=../src/network/model/packet-tag-list.cc, line=250
terminate called without an active exception
Command ['/home/ns3_project/ns-allinone-3.26/ns-3.26/build/src/aomdv/examples/ns3.26-aomdv-example-debug'] terminated with signal SIGIOT. Run it under a debugger to get more information (./waf --run <program> --command-template="gdb --args %s <args>").
root@fe6cae3aeef0:/home/ns3_project/ns-allinone-3.26/ns-3.26#

[Gabriel Ferreira]

ns-3.26 is 10 years old. Are you sure you need to use that particular version? 

[eyasu tadese]

yes

[Tommaso Pecorella]

Let me rephrase what Gabriel said.

1. ns-3.26 is 10-years old, and we can't offer reliable support (or even fully meaningful suggestions) on it.
2. aomdv was (is) an external contributed model, so the maintainers never had the chance to fully comb though it.

Said so, I can give you a hint, but not a solution.

Packet Tags and Byte Tags are nifty solutions to add metadata to a packet - something that inside an OS is usually very common. Of course they're lost when you actually transmit a packet, so using a tag that the sender did add to the packet on reception is definitely a bad idea. You can use them in a simulation to simplify some operations, but they're not something that a real protocol could use.

What is (probably) happening here is that AOMDV is adding a tag when you transmit a packet and that tag is not removed before sending the data on the wire. Perhaps it's even used by intermediate nodes or on reception. Something that I would definitely classify as a no-no.

Ping takes the packet it receives and turns it to a reply, so if the tag has not been removed, you'll end up with a trying to transmit a packet whose data has been already tagged, and a crash.

The solution might range from a dirty fix (remove the tags before passing the packet to the upper layers) to a more serious question: is the AOMDV model using tricks that makes it so unrealistic that its results are unreliable?

SO, the question leads back to what Gabriel asked (a little bit expanded): are you sure you want to use a 10-years old ns-3 and a model that has some red flags waving around?

In both cases, we're sorry but we can't help you more than this.