Bug 2854

Summary:	he-wifi-network crashes when frequency is set at 2.4 GHz
Product:	ns-3	Reporter:	sebastien.deronne
Component:	wifi	Assignee:	sebastien.deronne
Status:	RESOLVED FIXED
Severity:	normal	CC:	ns-bugs, redieteab.orange
Priority:	P3
Version:	ns-3-dev
Hardware:	All
OS:	All
Bug Depends on:
Bug Blocks:	2853
Attachments:	patch patch Allow 40 MHz at 5GHz allow 40 MHz at 5GHz

Description sebastien.deronne 2018-01-13 05:31:54 UTC

./waf --run "he-wifi-network --simulationTime=0.25 --frequency=2.4 --useRts=0"

MCS value		Channel width		GI			Throughput
0			20 MHz			3200 ns			6.21773 Mbit/s
0			20 MHz			1600 ns			6.92429 Mbit/s
0			20 MHz			800 ns			7.39533 Mbit/s
0			40 MHz			3200 ns			12.5768 Mbit/s
0			40 MHz			1600 ns			14.2254 Mbit/s
0			40 MHz			800 ns			15.0733 Mbit/s
0			80 MHz			3200 ns			12.9065 Mbit/s
Command ['/Users/Sebastien/ns-3/ns-3-allinone/ns-3-dev/build/examples/wireless/ns3-dev-he-wifi-network-debug', '--simulationTime=0.25', '--frequency=2.4', '--useRts=0'] exited with code 1


./waf --run "he-wifi-network --simulationTime=0.25 --frequency=2.4 --useRts=1"

MCS value		Channel width		GI			Throughput
0			20 MHz			3200 ns			6.02931 Mbit/s
0			20 MHz			1600 ns			6.68877 Mbit/s
0			20 MHz			800 ns			7.1127 Mbit/s
0			40 MHz			3200 ns			12.247 Mbit/s
0			40 MHz			1600 ns			13.8486 Mbit/s
0			40 MHz			800 ns			14.6493 Mbit/s
0			80 MHz			3200 ns			12.5768 Mbit/s
Command ['/Users/Sebastien/ns-3/ns-3-allinone/ns-3-dev/build/examples/wireless/ns3-dev-he-wifi-network-debug', '--simulationTime=0.25', '--frequency=2.4', '--useRts=1'] exited with code 1

Comment 1 sebastien.deronne 2018-01-13 06:47:21 UTC

It crashes because throughput at 80 MHz is too low and so it does not pass the check (no idea why the error message is not printed...).

Comment 2 sebastien.deronne 2018-01-13 07:04:59 UTC

Issue is caused by HtCapabilities which set the max supported channel width to 40. But my guess is that at 2.4 GHz the maximum allowed channel width is 40 MHz, so I suggest to change the example to only test 20 and 40 MHz channels at 2.4 GHz.

Tom, can you maybe confirm that 802.11ax is 20-40 MHz at 2.4 GHz, and 20-40-80-160 at 5 GHz?

Comment 3 sebastien.deronne 2018-01-13 09:43:01 UTC

Created attachment 3003 [details]
patch

If fine, I will push this.

Comment 4 sebastien.deronne 2018-01-13 10:50:06 UTC

Created attachment 3004 [details]
patch

Updated regression

Comment 5 sebastien.deronne 2018-01-24 13:11:35 UTC

Pushed in changeset 13264:c2f702e4adff

Comment 6 Rediet 2018-01-25 04:06:45 UTC

Sébastien, sorry to ask this question only now but why should 40 MHz be forbidden for HE in the 5GHz band?
If we compare 11ac behavior wrt 11n in the 5 GHz band, we can see that channel width is obtained from information contained in both HT Operation and VHT Operation (as detailed in section 11.40.1, and namely Table 11-24 "VHT BSS bandwidth", of 802.11-2016). The logic with Operation fields is also adopted by Capabilities fields (as implied in Table 9-249, which refers to section )
Don't you think that it would more logical that such a behavior would be adopted for 11ax?

Comment 7 sebastien.deronne 2018-01-25 07:14:40 UTC

Rediet, thanks, I think you are right, I closed this bug a bit too fast.
At 2.4 GHz, it should make use of both HT and HE capabilities. I will work on this later today if I have some time.

Comment 8 sebastien.deronne 2018-01-25 14:29:31 UTC

But still according to those capabilities, 40 MHz is the maximum at 2.4GHz, is my understanding correct?

Comment 9 sebastien.deronne 2018-01-25 14:49:06 UTC

Created attachment 3014 [details]
Allow 40 MHz at 5GHz

I reworked what I did so that we can now configure 40 MHz channel width at 5GHz, it is now looking at both HE and VHT capabilities to correctly set the supported channel width for a given peer.

Comment 10 sebastien.deronne 2018-01-25 16:23:22 UTC

Created attachment 3015 [details]
allow 40 MHz at 5GHz

Updated regression test

Comment 11 Rediet 2018-01-26 03:16:20 UTC

(In reply to sebastien.deronne from comment #8)
> But still according to those capabilities, 40 MHz is the maximum at 2.4GHz,
> is my understanding correct?

I'm not sure whether capabilities will be the limiting factor. It would rather be the regulatory per-region operating classes that would limit it to 40 MHz max. The operating class in the U.S. for the 2.4 GHz band forbids the use of channel 13, restricting the maximum band to 3 non-overlapping 20 MHz channels. However, in other regions, it would be theoretically possible to have a single 80 MHz channel covering the 2.4 GHz band (even if a practical use case can be hard to find, except when considering a deployment scenario in an isolated house somewhere in the Alps :-)).
Does this logic seem correct to you?

Comment 12 sebastien.deronne 2018-01-26 05:59:54 UTC

(In reply to Rediet from comment #11)
> (In reply to sebastien.deronne from comment #8)
> > But still according to those capabilities, 40 MHz is the maximum at 2.4GHz,
> > is my understanding correct?
> 
> I'm not sure whether capabilities will be the limiting factor. It would
> rather be the regulatory per-region operating classes that would limit it to
> 40 MHz max. The operating class in the U.S. for the 2.4 GHz band forbids the
> use of channel 13, restricting the maximum band to 3 non-overlapping 20 MHz
> channels. However, in other regions, it would be theoretically possible to
> have a single 80 MHz channel covering the 2.4 GHz band (even if a practical
> use case can be hard to find, except when considering a deployment scenario
> in an isolated house somewhere in the Alps :-)).
> Does this logic seem correct to you?

Yes, regulations should for sure always apply.
In any case, I think we can now close this bug with the updated patch that allows again 40 MHz.

Comment 13 sebastien.deronne 2018-01-31 14:19:29 UTC

Modifications pushed in changeset 13269:aec4048edcac