[Ns-developers] Compile error when setting up trace in phy-80211.cc

[Mathieu Lacage]

hi francesco,

[please, CC ns-developers with your patches next time: I think it is
nice when the other developers can monitor the state of development of
the 802.11 code.]

On Mon, 2006-01-16 at 03:10 +0100, Francesco Gringoli wrote:

> well, I attach a simple patch that
> 
> 
> - starts the timer m_txopStart also for the first station
> - does a simple check to not start timer m_txopStart (for the first
> and the others too) if the CAP is finished (it should happen...)

It would be nice to consolidate the code in a single location (in
startCurrentTxop for example) rather than duplicate it twice.


> I attach also a simple tcl script that simulates three stations. You
> can see that if you apply the patch I'm sending and switch off the
> patch that you gave me to let QSTAs sending QosNull there is still a
> misbehavior. Basically QSTA 3 will never be polled when data is
> available (after t=3.1s) because the timer m_txopStart that let the
> qap scheduler to switch from a QSTA to the next is too big so that
> there is not  time to perform TXOP for QSTA 3 inside the CAP. In fact
> 
> 
> delay=TxopDuration+ parameters ()->getPIFS ();

The PIFS is here to account for the PIFS delay which is required between
two TXOPs so it seems right to me to include it.

> 
> 
> and
> 
> 
> TotalCapTime = N*TxopDuration
> 
> 
> where N is the number of QSTAs; the last station will never be polled
> because of the PIFS added every time to the delay.
> I think that the delay used for this timer (which I just copied for
> the patch) should not taken into account the PIFS. Don't you?

maybe you need to add a PIFS in other places instead.

thanks for your efforts,
Mathieu
--