Throughput unfairness and effect of CCAMode1Threshold in simple terms

[Mahesh]

Hi all,

I am running a simulation in which I have a static network of wifi nodes connected through p2p to a csma backhaul on ns-3.26. I am facing a significant difference in throughput among various flows where wifi nodes are sending cbr traffic to csma backhaul. 

Also, when I change the CCAMode1Threshold, the results are changing a lot. How to choose CCAMode1Threshold and why is there a big difference in throughput among various flows? Each flow is started with a difference of 1 second.

The same script is giving the following error on ns3-dev:

msg="Failed to bind socket", file=../src/applications/model/packet-sink.cc, line=119

terminate called without an active exception

Why is this happening? Looking forward to your kind suggestions. Script is attached herewith along with a sample output where 3 flows are simulated.

Thanks,

Mahesh

[Mahesh]

Kindly reply.

[Tommaso Pecorella]

Read the posting guidelines. Soliciting a reply is dramatically rude.

T.

On Saturday, February 24, 2018 at 8:28:36 AM UTC-6, Mahesh wrote:

Kindly reply.

[Tommaso Pecorella]

Moreover:

1) Please fix your own coding mistakes, I can't even compile your code:

../scratch/isotropic-hybrid-16-nodes-dev.cc:495:20: error: variable-sized object may not be initialized

  int source_nodes[m_numNodes] = {0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15};

                   ^~~~~~~~~~

1 error generated.

2)

      ac_sinkTcp[i] = sinkTcp.Install (nc_csma.Get(1));

Installing an application TWICE or more on the same node, and asking the applications to use the same port is definitely a bad idea. I'm not surprised that you get run-time errors.

T.