[ns] 802.11 MAC/PHY questions

["Neeraj Poojary" <legolas_3@hotmail.com>]

hi all,
  i modified the wireless script simple-wireless.tcl, from Marc Greis' 
tutorial, to have four ftp connections (eight nodes) and no mobility.
the nodes are close enough to not lose packets due to distance. i had
three of the ftp connections starting at the same instant (t=13). i'm
tracing the tcp variables ndatapack_ and nrexmitpack_ for each agent.
each tracefile is exactly identical (except for node id) and looks like 
this:

^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
f t0 a_o142 nndatapack_ v0
f t0 a_o142 nnrexmitpack_ v0
f t13 a_o142 nndatapack_ v1
f t19 a_o142 nndatapack_ v2
f t19 a_o142 nnrexmitpack_ v1
f t31 a_o142 nndatapack_ v3
f t31 a_o142 nnrexmitpack_ v2
f t55 a_o142 nndatapack_ v4
f t55 a_o142 nnrexmitpack_ v3
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

the retransmits seem to be occurring at 6, 12, 24,... second intervals, but 
at exactly the same instant at each node each time, so that they keep 
bumping heads and not one packet gets across. i'm assuming the 
retransmissions are due to collisions. this begs two questions:

1> shouldn't the backoffs be randomized to avoid this situation?
2> the NS manual says the PHY layer approximates spread spectrum (DSSS). in 
that case shouldn't some of the packets get across some of the time, even if 
they 'collide'?

finally, a third question would be
3> how can i find out the reason for a retransmit (collision/ACK 
timeout...)?

apart from increasing the number of nodes and connections, making the
nodes stationary and adding the variable traces, i haven't changed anything 
in the Marc Greis script. things work better when i stagger
the start times of the three connections.

i'm new to NS. so its possible i'm doing something trivially wrong. any and 
all suggestions would be helpful.

thanks,
neeraj
______________________________________________________
Get Your Private, Free Email at http://www.hotmail.com