Wifi: log collision

[francesco]

Hi,

Is there a predefined tracepoint where I should attach to log wifi
collisions? If not, where should I create one?

Thanks,

Francesco

[francesco]

Ok, let me try again ;)

Would src/devices/wifi/dca-txop.cc:457 (in correspondence to "MY_DEBUG
("collision")") be a good idea?

[Mathieu Lacage]

On Tue, 2009-02-17 at 00:01 -0800, francesco wrote:
> Ok, let me try again ;)

opps, I missed your first email.

> Would src/devices/wifi/dca-txop.cc:457 (in correspondence to "MY_DEBUG
> ("collision")") be a good idea?

Well, it really depends on what you define as being a collision but this
debug statement will be triggered when the user has just requested
access to the medium, that the backoff count has reached zero and that
the medium is busy.

regards,
Mathieu

[francesco]

Thanks for the answer!

> Well, it really depends on what you define as being a collision but this
> debug statement will be triggered when the user has just requested
> access to the medium, that the backoff count has reached zero and that
> the medium is busy.

I am just interested in an indication that "the medium is becoming
crowdy", so I guess that that debug statement would be enough.
Maybe you can suggest a more meaningful name (i.e. than just
"Collision") for the trace source, so that it can be more generally
useful?

[Mathieu Lacage]

On Tue, 2009-02-17 at 01:00 -0800, francesco wrote:

> > Well, it really depends on what you define as being a collision but this
> > debug statement will be triggered when the user has just requested
> > access to the medium, that the backoff count has reached zero and that
> > the medium is busy.
>
> I am just interested in an indication that "the medium is becoming
> crowdy", so I guess that that debug statement would be enough.

A decent indicator of this is the percentage of busy time on the medium
which you could calculate from the PHY trace events. I remember seeing
papers which inferred stuff from that busy/idle ratio. Don't ask me
which.

Mathieu

[francesco]

Sorry, I was not clear, I meant "the medium is crowdy _and stations
have trouble accessing it". If there are just two hosts tranferring
some huge amount of tata the medium will be busy for most of the time,
but no (or few) "collisions" take place.

Btw, which is the trace-path of DcaTxop? It is not stated in the
doxygen online pages...

[Mathieu Lacage]

On Tue, 2009-02-17 at 01:27 -0800, francesco wrote:

> Btw, which is the trace-path of DcaTxop? It is not stated in the

There is none because m_dca is not declared as an attribute in
nqap-wifi-mac.cc, nqsta-wifi-mac.cc and adhoc-wifi-mac.cc

You need something like this (untested):

.AddAttribute ("DcaTxop", "XXX",
PointerValue (),
MakePointerAccessor (&NqstaWifiMac::DoGetDcaTxop),
MakePointerChecker<DcaTxop> ())

regards,
Mathieu

[francesco]

I have opted for MediumBusy as a trace name.
http://www.nsnam.org/bugzilla/show_bug.cgi?id=500