Bug 2813 – OFDM 10 MHz and 5 MHz PHYs do not have correct subcarrier frequency spacings

Bug 2813 - OFDM 10 MHz and 5 MHz PHYs do not have correct subcarrier frequency spacings


Summary:	OFDM 10 MHz and 5 MHz PHYs do not have correct subcarrier frequency spacings

Status:	RESOLVED FIXED

Product:	ns-3
Classification:	Unclassified
Component:	wifi
Version:	ns-3.27
Hardware:	All All

Importance:	P3 normal
Assigned To:	Tom Henderson

URL:
Keywords:

Depends on:
Blocks:
	Show dependency tree / graph

Reported:	2017-11-02 06:02 UTC by Rediet
Modified:	2018-02-20 07:42 UTC (History)
CC List:	3 users (show)

See Also:

Attachments
proposed patch (17.30 KB, patch) 2018-02-18 11:51 UTC, Tom Henderson	Details \| Diff
simple test program (1.32 KB, text/x-c++src) 2018-02-18 11:52 UTC, Tom Henderson	Details
Add an attachment (proposed patch, testcase, etc.)

Note You need to log in before you can comment on or make changes to this bug.

Description Rediet 2017-11-02 06:02:04 UTC

IEEE standard specifies that OFDM PHYs with 10 MHz and 5 MHz channel spacing should respectively use 156.25 kHz and 78.125 kHz subcarrier frequency spacing. In the current state of ns-3, the default 312.5 kHz spacing is being used (except for 11ax). With the introduction of Spectrum framework in wifi module, the way subcarriers are modeled becomes critical. It would thus be more realistic to use the correct subcarrier spacings (and subsequent FFT periods).

Comment 1 sebastien.deronne 2017-11-20 16:45:58 UTC

Tom, I guess you are working on this?

Comment 2 Tom Henderson 2017-11-20 16:47:41 UTC

Yes.

Comment 3 Tom Henderson 2018-02-18 11:51:47 UTC

Created attachment 3047 [details]
proposed patch

This patch corrects the spectrum value layout for 10 MHz and 5 MHz channels.  The number of bands per channel is the same, but the band bandwidth is reduced with channel width.  The guard band around the channel is also reduced in a corresponding manner.  The longer symbol duration is handled elsewhere in the code and didn't seem to need to be modified.  This patch doesn't address changes to CCA/ED thresholds for smaller channels (there is another open bug to clean up the CCA behavior and attributes).  Python bindings would need to be rescanned.

Comment 4 Tom Henderson 2018-02-18 11:52:50 UTC

Created attachment 3048 [details]
simple test program

This program can be run from scratch/ to display the resulting spectrum value for 20, 10, and 5 MHz channels.

Comment 5 sebastien.deronne 2018-02-18 11:59:12 UTC

+1 for me, thanks Tom.

Comment 6 Rediet 2018-02-19 04:29:17 UTC

Thanks Tom. +1 for me too.

Comment 7 sebastien.deronne 2018-02-19 04:31:34 UTC

I think we can push this one then :-)

Comment 8 sebastien.deronne 2018-02-20 07:42:44 UTC

Pushed in changeset 13298:23aa339aecd6