Channel scanning in 802.11

[Ramon Bauza]

Hi all,

I have seen in the ns-3 manual, that although WiFi networks can be
simulated in infraestructure mode, the current ns-3 version does not
perfom channel scanning. My questions are then:

- Can a NetDevice switch among different channels at execution time?
- Is the channel scanning going to be implemented?

Many thanks,
Ramon.

[Mathieu Lacage]

On Tue, 2008-10-07 at 04:23 -0700, Ramon Bauza wrote:
> Hi all,
>
> I have seen in the ns-3 manual, that although WiFi networks can be
> simulated in infraestructure mode, the current ns-3 version does not
> perfom channel scanning. My questions are then:
>
> - Can a NetDevice switch among different channels at execution time?

You can change the frequency band used by a device but there is no
support in the current PHY model to modelize interference among signals
in overlapping frequency bands. A patch/model description to do so would
be welcome.

> - Is the channel scanning going to be implemented?

yes, if you send a patch.

regards,
Mathieu

[Gustavo Carneiro]

A while back I had written some wifi scanning code.  I have re-synced with ns-3-dev and uploaded the code here, in case it is useful to someone:

http://code.nsnam.org/gjc/ns-3-wifi-scanning/

Caveats:

  1- The code is experimental;

  2- The active scanning is pessimistic, i.e. scanning takes longer time than real world, because it assumes all channels are busy with traffic and always spends 50 ms in each channel (real world cards are smarter and spend just a few ms if they do not sense any traffic in that channel).  In any case, it works, in spite of being rudimentary.

 3- The code has not been reviewed yet by Mathieu, the WiFi maintainer, although I hope he will consider it ;-)

 4- I don't have a lot of time to maintain the code, so it would be wonderful if someone could pick it up.  Maybe I might develop something more on top of this, or maybe not, I am not sure yet...

Further things to implement:

  1- Switching channels should consume between 1 and 5 ms (hardware limitations) [1];

  2- Probing a "free" channel should require a delay of only 1-2 ms (aka "minimum channel time"), and 10 to 27 ms (aka "maximum channel time") for busy channels [1];

  3- Due to inter-channel interference, additional channels may erroneously appear busy and thus cause additional scanning delays (has to spend maximum channel time, instead of minimum, for additional channels), so without the inter-channel interference model mentioned by Mathieu scanning delays might be too optimistic;

  4- The scanning machinery should somehow record the "signal quality" of the received ProbeResponses, and report it back to the client, so that the client can choose the best AP based on signal strength;

  5- The final piece of the puzzle would be to implement a MobilityManager kind of object, attached to the node, which would select the best AP at all times.  Right now NS 3 has very rudimentary mobility management.  An experiment I have made: place two APs on the same channel, same SSID, spaced by 100 meters;  move a STA to the space in the middle of the two APs; the STA starts receiving beacons from both APs, and it keeps associating to one and other AP, constantly flip-flopping its association between the two APs.

[1] Murray, D.   Dixon, M.   Koziniec, T., "Scanning Delays in 802.11 Networks", NGMAST '07. (http://ieeexplore.ieee.org/xpls/abs_all.jsp?arnumber=4343430)

2008/10/7 Mathieu Lacage <mathieu...@sophia.inria.fr>

--
Gustavo J. A. M. Carneiro
INESC Porto, Telecommunications and Multimedia Unit
"The universe is always one step beyond logic." -- Frank Herbert

[mathieu lacage]

On Fri, 2008-10-10 at 15:16 +0000, Giuseppe Barba wrote:
> Hi all,
> in order to correctly simulate scenarios with multi channel communication,
> you must model the adjacent channel interference (ACI) and the adjacent
> channel rejection (ACR). For this reason is necessary to preventively
> engineering the Phy model.
> For this scope we can just consider the ideal TX power masks of various
> classes of TX filter (class A, B, C e D) to model the overlapping channel
> (ACI), and the RX filter mask to consider the rejection to adjacent
> interference with various classes of filter (e.g. class 2).
> For example: if a node transmit over an adjacent channel, we can compute the
> correspondent overlapped TX interfered power by attenuate the TX power by a
> factor depending on Tx mask at frequency shift. Then we can compute the RX
> interfering power through loss model adopted for in channel communication.

Yes, that is pretty similar to what is done in nsmiracle and, yes, I
would be happy to see a patch to implement something similar :)

Mathieu

[Ayinebyona Eliab]

Hey Gustavo, Am working on something related to channel scanning. would you please give me your code for reference. I checked the link that you gave and nothing is there. Thanks.

[Tommaso Pecorella]

You just replied to a mail that is 7 years old. Congratulations.