[Ns-developers] NS-3 Wifi contributions

[Mathieu Lacage]

hi federico,

On Fri, 2008-03-21 at 16:26 +0100, Federico Maguolo wrote:

> I think that a better solution is to introduce two new methods in the 
> WifiRemoteStation class:
> -NeedRtsRetransmission
> -NeedDataRetransmission
> This two methods are used when a Rts frame or a data frame fail. If you agree 
> with me I've already done a patch which modify the objects. You can find this 
> patch at the usual link

I think you managed to read what I had in mind despite the bad
description I gave.

> 
> http://www.dei.unipd.it/wdyn/index.php?IDsezione=5517
> 
> in the section "New retransmission interface for the WifiRemoteStation".

I believe that the m_ssrc/m_slrc counters should be incremented by
ReportRtsFailed,  ReportDataFailed, ReportRtsOk, and ReportDataFailed
(all of which are invoked by MacLow) instead of being incremented by
NeedDataTransmission/NeedRtsTransmission. And, you should reset them
from ReportRtsFinalFailed and ReportDataFinalFailed which are invoked
from DcaTxop::MissedCts and DcaTxop::MissedData.

You will need to be careful about the order in which these methods are
invoked. It might be a great idea to actually write some documentation
for these above methods (which would include a description of the
expected call ordering) in
src/devices/wifi/wifi-remote-station-manager.h. Yes, I know that I
should have done that myself.

> Moreover, dealing  with this modification I found a problem the default value 
> of a WifiMode has to be set: When you do not specify an explicit value to the 
> attribute, the default value could be set when an object is created. I 
> discovered that it is not true for the DataMode and the CtlMode of the 
> ConstantRateWifiManager. To understand what I mean I suggest to try to run 
> this example:
> 
> http://www.dei.unipd.it/~maguolo/example.cc

Can you check that the current tree does not have this problem anymore ?

> In the section "Initatial value problem for WifiMode" at the address 
> http://www.dei.unipd.it/wdyn/index.php?IDsezione=5517,  you could find a fix 
> proposal, but I don't know if there are a better ways. Moreover, the patch 
> that you find here integrates the bugfix on the WifiMode, because I saw that 
> the AllocateUid method has the same problem.

The Object part of the patch above has already been fixed in trunk so,
you probably should update your tree. The AllocateUid method seemed to
be buggy indeed. I think I understand why it worked despite being buggy
but I have a hard time figuring out how it could fail. Anyway, I have
pushed a version which is hopefully more robust: if it still fails,
please, let me know.

Mathieu