[ns] Simulating Hidden Nodes with LANs

[Iain Marcuson]

I am trying to simulate a network divided in four main groups.  The nodes at the edges between groups belong to both groups.  These create the hidden nodes.

The first two groups look something like this


o         ooo         *  *       xxx       x

Here, the "o"'s belong to one group, the "x"'s to another, and the "*"'s bridge between both (these bridging nodes share  both groups) and thus can be susceptible to the hidden node problem.

>From searching the mailing lists for "hidden node", I have seen that 802.11 can simulate this when the MAC scheme is changed from a RTS/CTS scheme to a CSMA scheme.  However, for one of the scenarios I plan to study, I need to make large changes to the topology of the network.  Given the groups

1    2    3    4

where a group can only directly reach a neighbor, I want to create a bridge that completely links Group 1 to Group 3, simulating Sporadic E Propagation.  However, from what I have seen of the parameters for wireless networks, the wireless networks seem to be geographically-oriented as opposed to topographically-oriented, which suggests that I cannot create the desired topology using the wireless links.  LANs can have an arbitrary topology, but I have not seen how to "merge" their MAC layers, needed for simulating lost packets at the nodes bridging groups.

Searching the mailing list for "node collision" and "inter lan" did not yield an answer.  I am using ns-2.33, Tcl/Tk 8.4, compiled from source on LFS-6.3.  Thank you for your consideration.