[ath9k-devel] why is bitrate of 65.0 MBit/s MCS 7 appropriate for signal strength of -93 dBm?

[ath9k-devel] why is bitrate of 65.0 MBit/s MCS 7 appropriate for signal strength of -93 dBm?

[Larry Vaden]

Mindful that we are all ignorant, just about different things, I ask
the ? in the subject line.

Would that not influence the success ratio and the performance?

AP (Ubiquiti Rocket M900, running stock r26496):

root at OpenWrt:~# iw wlan0 station dump
Station 00:15:6d:9c:16:df (on wlan0)
        inactive time:  3620 ms
        rx bytes:       1747
        rx packets:     42
        tx bytes:       3034
        tx packets:     45
        tx retries:     308
        tx failed:      23
        signal:         -93 dBm
        signal avg:     -105 dBm
        tx bitrate:     65.0 MBit/s MCS 7
        rx bitrate:     6.5 MBit/s MCS 0


root at OpenWrt:~# iw wlan0 station dump
Station 00:15:6d:9c:16:df (on wlan0)
        inactive time:  2660 ms
        rx bytes:       323
        rx packets:     11
        tx bytes:       616
        tx packets:     9
        tx retries:     49
        tx failed:      3
        signal:         -93 dBm
        signal avg:     -96 dBm
        tx bitrate:     6.5 MBit/s MCS 0
        rx bitrate:     6.5 MBit/s MCS 0
root at OpenWrt:~#

CPE (Ubiquiti Loco M900, running stock r26496):

root at OpenWrt:~# iw wlan0 station dump
Station 00:15:6d:4e:f1:46 (on wlan0)
	inactive time:	6950 ms
	rx bytes:	68637
	rx packets:	907
	tx bytes:	2373
	tx packets:	24
	tx retries:	88
	tx failed:	3
	signal:  	-101 dBm
	signal avg:	-100 dBm
	tx bitrate:	6.5 MBit/s MCS 0
	rx bitrate:	39.0 MBit/s MCS 4
root at OpenWrt:~#

best regards/ldv
-- 
Larry Vaden, CoFounder
Internet Texoma, Inc.
Serving Rural Texomaland Since 1995
We Care About Your Connection!

[ath9k-devel] why is bitrate of 65.0 MBit/s MCS 7 appropriate for signal strength of -93 dBm?

[Mohammed Shafi]

On Wed, Apr 6, 2011 at 3:28 PM, Larry Vaden <vaden@texoma.net> wrote:
> Mindful that we are all ignorant, just about different things, I ask
> the ? in the subject line.
>
> Would that not influence the success ratio and the performance?

ath9k rate control is based on PER(packet error rate), but previously
it is based on RSSI.
But i doubt you are using minstrel rate control, for that please read
http://wireless.kernel.org/en/developers/Documentation/mac80211/RateControl/minstrel

>
> AP (Ubiquiti Rocket M900, running stock r26496):
>
> root at OpenWrt:~# iw wlan0 station dump
> Station 00:15:6d:9c:16:df (on wlan0)
> ? ? ? ?inactive time: ?3620 ms
> ? ? ? ?rx bytes: ? ? ? 1747
> ? ? ? ?rx packets: ? ? 42
> ? ? ? ?tx bytes: ? ? ? 3034
> ? ? ? ?tx packets: ? ? 45
> ? ? ? ?tx retries: ? ? 308
> ? ? ? ?tx failed: ? ? ?23
> ? ? ? ?signal: ? ? ? ? -93 dBm
> ? ? ? ?signal avg: ? ? -105 dBm
> ? ? ? ?tx bitrate: ? ? 65.0 MBit/s MCS 7
> ? ? ? ?rx bitrate: ? ? 6.5 MBit/s MCS 0
>
>
> root at OpenWrt:~# iw wlan0 station dump
> Station 00:15:6d:9c:16:df (on wlan0)
> ? ? ? ?inactive time: ?2660 ms
> ? ? ? ?rx bytes: ? ? ? 323
> ? ? ? ?rx packets: ? ? 11
> ? ? ? ?tx bytes: ? ? ? 616
> ? ? ? ?tx packets: ? ? 9
> ? ? ? ?tx retries: ? ? 49
> ? ? ? ?tx failed: ? ? ?3
> ? ? ? ?signal: ? ? ? ? -93 dBm
> ? ? ? ?signal avg: ? ? -96 dBm
> ? ? ? ?tx bitrate: ? ? 6.5 MBit/s MCS 0
> ? ? ? ?rx bitrate: ? ? 6.5 MBit/s MCS 0
> root at OpenWrt:~#
>
> CPE (Ubiquiti Loco M900, running stock r26496):
>
> root at OpenWrt:~# iw wlan0 station dump
> Station 00:15:6d:4e:f1:46 (on wlan0)
> ? ? ? ?inactive time: ?6950 ms
> ? ? ? ?rx bytes: ? ? ? 68637
> ? ? ? ?rx packets: ? ? 907
> ? ? ? ?tx bytes: ? ? ? 2373
> ? ? ? ?tx packets: ? ? 24
> ? ? ? ?tx retries: ? ? 88
> ? ? ? ?tx failed: ? ? ?3
> ? ? ? ?signal: ? ? ? ? -101 dBm
> ? ? ? ?signal avg: ? ? -100 dBm
> ? ? ? ?tx bitrate: ? ? 6.5 MBit/s MCS 0
> ? ? ? ?rx bitrate: ? ? 39.0 MBit/s MCS 4
> root at OpenWrt:~#
>
> best regards/ldv
> --
> Larry Vaden, CoFounder
> Internet Texoma, Inc.
> Serving Rural Texomaland Since 1995
> We Care About Your Connection!
> _______________________________________________
> ath9k-devel mailing list
> ath9k-devel at lists.ath9k.org
> https://lists.ath9k.org/mailman/listinfo/ath9k-devel
>

[ath9k-devel] why is bitrate of 65.0 MBit/s MCS 7 appropriate for signal strength of -93 dBm?

[Larry Vaden]

On Wed, Apr 6, 2011 at 5:26 AM, Mohammed Shafi <shafi.ath9k@gmail.com> wrote:
> On Wed, Apr 6, 2011 at 3:28 PM, Larry Vaden <vaden@texoma.net> wrote:
>> Mindful that we are all ignorant, just about different things, I ask
>> the ? in the subject line.
>>
>> Would that not influence the success ratio and the performance?
>
> ath9k rate control is based on PER(packet error rate), but previously
> it is based on RSSI.
> But i doubt you are using minstrel rate control, for that please read
> http://wireless.kernel.org/en/developers/Documentation/mac80211/RateControl/minstrel

With all due respect, I do not think your answer is responsive to the
question and the data presented.

>> ? ? ? ?tx packets: ? ? 45
>> ? ? ? ?tx retries: ? ? 308
>> ? ? ? ?tx failed: ? ? ?23

Note a FAILURE RATE of 50% and retries ~ 700%.

So, why is 65.0 MBit/s MCS 7 even on the table if rate control is based on PER?

Is it possible that some code other than ath9k is jacking with the
data rates (read: I am ignorant about what protections OpenWrt might
have)?

regards/ldv

>> AP (Ubiquiti Rocket M900, running stock r26496):
>>
>> root at OpenWrt:~# iw wlan0 station dump
>> Station 00:15:6d:9c:16:df (on wlan0)
>> ? ? ? ?inactive time: ?3620 ms
>> ? ? ? ?rx bytes: ? ? ? 1747
>> ? ? ? ?rx packets: ? ? 42
>> ? ? ? ?tx bytes: ? ? ? 3034
>> ? ? ? ?tx packets: ? ? 45
>> ? ? ? ?tx retries: ? ? 308
>> ? ? ? ?tx failed: ? ? ?23
>> ? ? ? ?signal: ? ? ? ? -93 dBm
>> ? ? ? ?signal avg: ? ? -105 dBm
>> ? ? ? ?tx bitrate: ? ? 65.0 MBit/s MCS 7
>> ? ? ? ?rx bitrate: ? ? 6.5 MBit/s MCS 0
>>
>>
>> root at OpenWrt:~# iw wlan0 station dump
>> Station 00:15:6d:9c:16:df (on wlan0)
>> ? ? ? ?inactive time: ?2660 ms
>> ? ? ? ?rx bytes: ? ? ? 323
>> ? ? ? ?rx packets: ? ? 11
>> ? ? ? ?tx bytes: ? ? ? 616
>> ? ? ? ?tx packets: ? ? 9
>> ? ? ? ?tx retries: ? ? 49
>> ? ? ? ?tx failed: ? ? ?3
>> ? ? ? ?signal: ? ? ? ? -93 dBm
>> ? ? ? ?signal avg: ? ? -96 dBm
>> ? ? ? ?tx bitrate: ? ? 6.5 MBit/s MCS 0
>> ? ? ? ?rx bitrate: ? ? 6.5 MBit/s MCS 0
>> root at OpenWrt:~#
>>
>> CPE (Ubiquiti Loco M900, running stock r26496):
>>
>> root at OpenWrt:~# iw wlan0 station dump
>> Station 00:15:6d:4e:f1:46 (on wlan0)
>> ? ? ? ?inactive time: ?6950 ms
>> ? ? ? ?rx bytes: ? ? ? 68637
>> ? ? ? ?rx packets: ? ? 907
>> ? ? ? ?tx bytes: ? ? ? 2373
>> ? ? ? ?tx packets: ? ? 24
>> ? ? ? ?tx retries: ? ? 88
>> ? ? ? ?tx failed: ? ? ?3
>> ? ? ? ?signal: ? ? ? ? -101 dBm
>> ? ? ? ?signal avg: ? ? -100 dBm
>> ? ? ? ?tx bitrate: ? ? 6.5 MBit/s MCS 0
>> ? ? ? ?rx bitrate: ? ? 39.0 MBit/s MCS 4
>> root at OpenWrt:~#
>>
>> best regards/ldv
>> --
>> Larry Vaden, CoFounder
>> Internet Texoma, Inc.
>> Serving Rural Texomaland Since 1995
>> We Care About Your Connection!
>> _______________________________________________
>> ath9k-devel mailing list
>> ath9k-devel at lists.ath9k.org
>> https://lists.ath9k.org/mailman/listinfo/ath9k-devel
>>
>



-- 
Larry Vaden, CoFounder
Internet Texoma, Inc.
Serving Rural Texomaland Since 1995
We Care About Your Connection!

[ath9k-devel] why is bitrate of 65.0 MBit/s MCS 7 appropriate for signal strength of -93 dBm?

[Felix Fietkau]

On 2011-04-06 4:17 PM, Larry Vaden wrote:
> On Wed, Apr 6, 2011 at 5:26 AM, Mohammed Shafi<shafi.ath9k@gmail.com>  wrote:
>>  On Wed, Apr 6, 2011 at 3:28 PM, Larry Vaden<vaden@texoma.net>  wrote:
>>>  Mindful that we are all ignorant, just about different things, I ask
>>>  the ? in the subject line.
>>>
>>>  Would that not influence the success ratio and the performance?
>>
>>  ath9k rate control is based on PER(packet error rate), but previously
>>  it is based on RSSI.
>>  But i doubt you are using minstrel rate control, for that please read
>>  http://wireless.kernel.org/en/developers/Documentation/mac80211/RateControl/minstrel
>
> With all due respect, I do not think your answer is responsive to the
> question and the data presented.
>
>>>          tx packets:     45
>>>          tx retries:     308
>>>          tx failed:      23
>
> Note a FAILURE RATE of 50% and retries ~ 700%.
Simple packet/retry/fail counters are completely misleading when it 
comes to debugging wireless issues.

> So, why is 65.0 MBit/s MCS 7 even on the table if rate control is based on PER?
Because the assumption that lower rates always work better than higher 
rates is a bad one to make in a rate control algorithm.

> Is it possible that some code other than ath9k is jacking with the
> data rates (read: I am ignorant about what protections OpenWrt might
> have)?
You can find better statistics with much more meaningful output in
/sys/kernel/debug/ieee80211/phy0/netdev:wlan0/stations/*/rc_stats

- Felix

[ath9k-devel] why is bitrate of 65.0 MBit/s MCS 7 appropriate for signal strength of -93 dBm?

[Larry Vaden]

On Wed, Apr 6, 2011 at 9:31 AM, Felix Fietkau <nbd@openwrt.org> wrote:
>> So, why is 65.0 MBit/s MCS 7 even on the table if rate control is based on
>> PER?
>
> Because the assumption that lower rates always work better than higher rates
> is a bad one to make in a rate control algorithm.

I do not disagree.  I've seen plenty of cases where higher data rates
are more successful than lower data rates.

Is XR within the range of parameters tried by ath9k?

> You can find better statistics with much more meaningful output in
> /sys/kernel/debug/ieee80211/phy0/netdev:wlan0/stations/*/rc_stats

I'm not sure about the time domain covered by said stats as IIRC at
least 3 hours of 'iperf' has been run since r26494 was installed.

best regards/ldv

root at OpenWrt:~# cat
/sys/kernel/debug/ieee80211/phy0/netdev:wlan0/stations/*/rc_stats
type      rate     throughput  ewma prob   this prob  this
succ/attempt   success    attempts
HT20/LGI TtPMCS0        1.7       28.2       33.3          0(  0)
   34         193
HT20/LGI    MCS1        0.0        0.0        0.0          0(  0)
    0           0
HT20/LGI    MCS2        0.0        0.0        0.0          0(  0)
    0           0
HT20/LGI    MCS3        0.0        0.0        0.0          0(  0)
    0           1
HT20/LGI    MCS4        0.0        0.0        0.0          0(  0)
    0           0
HT20/LGI    MCS5        0.0        0.0        0.0          0(  0)
    0           1
HT20/LGI    MCS6        0.0        0.0        0.0          0(  0)
    0           0
HT20/LGI    MCS7        0.0        0.0        0.0          0(  0)
    0           1
HT20/LGI    MCS8        0.0        0.0        0.0          0(  0)
    0           0
HT20/LGI    MCS9        0.0        0.0        0.0          0(  0)
    0           0
HT20/LGI    MCS10       0.0        0.0        0.0          0(  0)
    0           0
HT20/LGI    MCS11       0.0        0.0        0.0          0(  0)
    0           1
HT20/LGI    MCS12       0.0        0.0        0.0          0(  0)
    0           0
HT20/LGI    MCS13       0.0        0.0        0.0          0(  0)
    0           0
HT20/LGI    MCS14       0.0        0.0        0.0          0(  0)
    0           1
HT20/LGI    MCS15       0.0        0.0        0.0          0(  0)
    0           1

Total packet count::    ideal 34      lookaround 6
Average A-MPDU length: 1.0
root at OpenWrt:~# uptime
 05:01:20 up  5:01, load average: 0.00, 0.01, 0.04
root at OpenWrt:~#


> - Felix
>



-- 
Larry Vaden, CoFounder
Internet Texoma, Inc.
Serving Rural Texomaland Since 1995
We Care About Your Connection!

[ath9k-devel] why is bitrate of 65.0 MBit/s MCS 7 appropriate for signal strength of -93 dBm?

[Felix Fietkau]

On 2011-04-06 4:49 PM, Larry Vaden wrote:
> On Wed, Apr 6, 2011 at 9:31 AM, Felix Fietkau<nbd@openwrt.org>  wrote:
>>>  So, why is 65.0 MBit/s MCS 7 even on the table if rate control is based on
>>>  PER?
>>
>>  Because the assumption that lower rates always work better than higher rates
>>  is a bad one to make in a rate control algorithm.
>
> I do not disagree.  I've seen plenty of cases where higher data rates
> are more successful than lower data rates.
>
> Is XR within the range of parameters tried by ath9k?
No. It's not even supported by the hardware.

>>  You can find better statistics with much more meaningful output in
>>  /sys/kernel/debug/ieee80211/phy0/netdev:wlan0/stations/*/rc_stats
>
> I'm not sure about the time domain covered by said stats as IIRC at
> least 3 hours of 'iperf' has been run since r26494 was installed.
>
> best regards/ldv
>
> root at OpenWrt:~# cat
> /sys/kernel/debug/ieee80211/phy0/netdev:wlan0/stations/*/rc_stats
> type      rate     throughput  ewma prob   this prob  this
> succ/attempt   success    attempts
> HT20/LGI TtPMCS0        1.7       28.2       33.3          0(  0)
>     34         193
> HT20/LGI    MCS1        0.0        0.0        0.0          0(  0)
>      0           0
> HT20/LGI    MCS2        0.0        0.0        0.0          0(  0)
>      0           0
> HT20/LGI    MCS3        0.0        0.0        0.0          0(  0)
>      0           1
> HT20/LGI    MCS4        0.0        0.0        0.0          0(  0)
>      0           0
> HT20/LGI    MCS5        0.0        0.0        0.0          0(  0)
>      0           1
> HT20/LGI    MCS6        0.0        0.0        0.0          0(  0)
>      0           0
> HT20/LGI    MCS7        0.0        0.0        0.0          0(  0)
>      0           1
> HT20/LGI    MCS8        0.0        0.0        0.0          0(  0)
>      0           0
> HT20/LGI    MCS9        0.0        0.0        0.0          0(  0)
>      0           0
> HT20/LGI    MCS10       0.0        0.0        0.0          0(  0)
>      0           0
> HT20/LGI    MCS11       0.0        0.0        0.0          0(  0)
>      0           1
> HT20/LGI    MCS12       0.0        0.0        0.0          0(  0)
>      0           0
> HT20/LGI    MCS13       0.0        0.0        0.0          0(  0)
>      0           0
> HT20/LGI    MCS14       0.0        0.0        0.0          0(  0)
>      0           1
> HT20/LGI    MCS15       0.0        0.0        0.0          0(  0)
>      0           1
>
> Total packet count::    ideal 34      lookaround 6
The total packet count indicates that there's no useful data in there 
(probably because a reconnect happened). The MCS7 rate that you saw was 
maybe just a probing attempt.

- Felix

[ath9k-devel] why is bitrate of 65.0 MBit/s MCS 7 appropriate for signal strength of -93 dBm?

[Larry Vaden]

On Wed, Apr 6, 2011 at 9:52 AM, Felix Fietkau <nbd@openwrt.org> wrote:
> On 2011-04-06 4:49 PM, Larry Vaden wrote:
>>
>> On Wed, Apr 6, 2011 at 9:31 AM, Felix Fietkau<nbd@openwrt.org> ?wrote:
>>>>
>>>> ?So, why is 65.0 MBit/s MCS 7 even on the table if rate control is based
>>>> on
>>>> ?PER?
>>>
>>> ?Because the assumption that lower rates always work better than higher
>>> rates
>>> ?is a bad one to make in a rate control algorithm.
>>
>> I do not disagree. ?I've seen plenty of cases where higher data rates
>> are more successful than lower data rates.
>>
>> Is XR within the range of parameters tried by ath9k?
>
> No. It's not even supported by the hardware.
>
>>> ?You can find better statistics with much more meaningful output in
>>> ?/sys/kernel/debug/ieee80211/phy0/netdev:wlan0/stations/*/rc_stats
>>
>> I'm not sure about the time domain covered by said stats as IIRC at
>> least 3 hours of 'iperf' has been run since r26494 was installed.
>>
>> best regards/ldv
>>
>> root at OpenWrt:~# cat
>> /sys/kernel/debug/ieee80211/phy0/netdev:wlan0/stations/*/rc_stats
>> type ? ? ?rate ? ? throughput ?ewma prob ? this prob ?this
>> succ/attempt ? success ? ?attempts
>> HT20/LGI TtPMCS0 ? ? ? ?1.7 ? ? ? 28.2 ? ? ? 33.3 ? ? ? ? ?0( ?0)
>> ? ?34 ? ? ? ? 193
>> HT20/LGI ? ?MCS1 ? ? ? ?0.0 ? ? ? ?0.0 ? ? ? ?0.0 ? ? ? ? ?0( ?0)
>> ? ? 0 ? ? ? ? ? 0
>> HT20/LGI ? ?MCS2 ? ? ? ?0.0 ? ? ? ?0.0 ? ? ? ?0.0 ? ? ? ? ?0( ?0)
>> ? ? 0 ? ? ? ? ? 0
>> HT20/LGI ? ?MCS3 ? ? ? ?0.0 ? ? ? ?0.0 ? ? ? ?0.0 ? ? ? ? ?0( ?0)
>> ? ? 0 ? ? ? ? ? 1
>> HT20/LGI ? ?MCS4 ? ? ? ?0.0 ? ? ? ?0.0 ? ? ? ?0.0 ? ? ? ? ?0( ?0)
>> ? ? 0 ? ? ? ? ? 0
>> HT20/LGI ? ?MCS5 ? ? ? ?0.0 ? ? ? ?0.0 ? ? ? ?0.0 ? ? ? ? ?0( ?0)
>> ? ? 0 ? ? ? ? ? 1
>> HT20/LGI ? ?MCS6 ? ? ? ?0.0 ? ? ? ?0.0 ? ? ? ?0.0 ? ? ? ? ?0( ?0)
>> ? ? 0 ? ? ? ? ? 0
>> HT20/LGI ? ?MCS7 ? ? ? ?0.0 ? ? ? ?0.0 ? ? ? ?0.0 ? ? ? ? ?0( ?0)
>> ? ? 0 ? ? ? ? ? 1
>> HT20/LGI ? ?MCS8 ? ? ? ?0.0 ? ? ? ?0.0 ? ? ? ?0.0 ? ? ? ? ?0( ?0)
>> ? ? 0 ? ? ? ? ? 0
>> HT20/LGI ? ?MCS9 ? ? ? ?0.0 ? ? ? ?0.0 ? ? ? ?0.0 ? ? ? ? ?0( ?0)
>> ? ? 0 ? ? ? ? ? 0
>> HT20/LGI ? ?MCS10 ? ? ? 0.0 ? ? ? ?0.0 ? ? ? ?0.0 ? ? ? ? ?0( ?0)
>> ? ? 0 ? ? ? ? ? 0
>> HT20/LGI ? ?MCS11 ? ? ? 0.0 ? ? ? ?0.0 ? ? ? ?0.0 ? ? ? ? ?0( ?0)
>> ? ? 0 ? ? ? ? ? 1
>> HT20/LGI ? ?MCS12 ? ? ? 0.0 ? ? ? ?0.0 ? ? ? ?0.0 ? ? ? ? ?0( ?0)
>> ? ? 0 ? ? ? ? ? 0
>> HT20/LGI ? ?MCS13 ? ? ? 0.0 ? ? ? ?0.0 ? ? ? ?0.0 ? ? ? ? ?0( ?0)
>> ? ? 0 ? ? ? ? ? 0
>> HT20/LGI ? ?MCS14 ? ? ? 0.0 ? ? ? ?0.0 ? ? ? ?0.0 ? ? ? ? ?0( ?0)
>> ? ? 0 ? ? ? ? ? 1
>> HT20/LGI ? ?MCS15 ? ? ? 0.0 ? ? ? ?0.0 ? ? ? ?0.0 ? ? ? ? ?0( ?0)
>> ? ? 0 ? ? ? ? ? 1
>>
>> Total packet count:: ? ?ideal 34 ? ? ?lookaround 6
>
> The total packet count indicates that there's no useful data in there
> (probably because a reconnect happened). The MCS7 rate that you saw was
> maybe just a probing attempt.
>
> - Felix

Read and understood.

[ath9k-devel] why is bitrate of 65.0 MBit/s MCS 7 appropriate for signal strength of -93 dBm?

[Larry Vaden]

On Wed, Apr 6, 2011 at 9:52 AM, Felix Fietkau <nbd@openwrt.org> wrote:
>>
>> Total packet count:: ? ?ideal 34 ? ? ?lookaround 6
>
> The total packet count indicates that there's no useful data in there
> (probably because a reconnect happened). The MCS7 rate that you saw was
> maybe just a probing attempt.

<learning moment, bitte>

What steps are necessary to learn about the reasons for the
disconnects/reconnects?

</learning moment, bitte>

ldv

[ath9k-devel] why is bitrate of 65.0 MBit/s MCS 7 appropriate for signal strength of -93 dBm?

[Mohammed Shafi]

On Thu, Apr 7, 2011 at 5:03 AM, Larry Vaden <vaden@texoma.net> wrote:
> On Wed, Apr 6, 2011 at 9:52 AM, Felix Fietkau <nbd@openwrt.org> wrote:
>>>
>>> Total packet count:: ? ?ideal 34 ? ? ?lookaround 6
>>
>> The total packet count indicates that there's no useful data in there
>> (probably because a reconnect happened). The MCS7 rate that you saw was
>> maybe just a probing attempt.
>
> <learning moment, bitte>
>
> What steps are necessary to learn about the reasons for the
> disconnects/reconnects?

*identifying the reason code for disconnection.
*enable mac80211 debugs
*looking at wpa-supplicant logs
*enabling ath9k debugging

>
> </learning moment, bitte>
>
> ldv
> _______________________________________________
> ath9k-devel mailing list
> ath9k-devel at lists.ath9k.org
> https://lists.ath9k.org/mailman/listinfo/ath9k-devel
>

[ath9k-devel] why is bitrate of 65.0 MBit/s MCS 7 appropriate for signal strength of -93 dBm?

[Mohammed Shafi]

On Thu, Apr 7, 2011 at 1:06 PM, Mohammed Shafi <shafi.ath9k@gmail.com> wrote:
> On Thu, Apr 7, 2011 at 5:03 AM, Larry Vaden <vaden@texoma.net> wrote:
>> On Wed, Apr 6, 2011 at 9:52 AM, Felix Fietkau <nbd@openwrt.org> wrote:
>>>>
>>>> Total packet count:: ? ?ideal 34 ? ? ?lookaround 6
>>>
>>> The total packet count indicates that there's no useful data in there
>>> (probably because a reconnect happened). The MCS7 rate that you saw was
>>> maybe just a probing attempt.
>>
>> <learning moment, bitte>
>>
>> What steps are necessary to learn about the reasons for the
>> disconnects/reconnects?
>
> *identifying the reason code for disconnection.
> *enable mac80211 debugs
> *looking at wpa-supplicant logs
> *enabling ath9k

also if its possible the sniffer capture can be highly helpful

>
>>
>> </learning moment, bitte>
>>
>> ldv
>> _______________________________________________
>> ath9k-devel mailing list
>> ath9k-devel at lists.ath9k.org
>> https://lists.ath9k.org/mailman/listinfo/ath9k-devel
>>
>

[ath9k-devel] why is bitrate of 65.0 MBit/s MCS 7 appropriate for signal strength of -93 dBm?

[Larry Vaden]

On Thu, Apr 7, 2011 at 2:36 AM, Mohammed Shafi <shafi.ath9k@gmail.com> wrote:
> On Thu, Apr 7, 2011 at 5:03 AM, Larry Vaden <vaden@texoma.net> wrote:
>> On Wed, Apr 6, 2011 at 9:52 AM, Felix Fietkau <nbd@openwrt.org> wrote:
>>>>
>>>> Total packet count:: ? ?ideal 34 ? ? ?lookaround 6
>>>
>>> The total packet count indicates that there's no useful data in there
>>> (probably because a reconnect happened). The MCS7 rate that you saw was
>>> maybe just a probing attempt.
>>
>> <learning moment, bitte>
>>
>> What steps are necessary to learn about the reasons for the
>> disconnects/reconnects?
>
> *identifying the reason code for disconnection.
> *enable mac80211 debugs
> *looking at wpa-supplicant logs
> *enabling ath9k debugging
>
>>
>> </learning moment, bitte>

Mohammed,

THANKS for your reply.

Please let this record be more detailed.  While I wrote my first line
of code in 1965, I am a n00b to OpenWrt and the fantastic ath9k and
have 64 year old neurons.

If I have missed a wiki article, that will suffice.

We also need to know if the AR9280 has a self-test which can be invoked.

Again, THANKS for your reply.

[ath9k-devel] why is bitrate of 65.0 MBit/s MCS 7 appropriate for signal strength of -93 dBm?

[Daniel Halperin]

On Thu, Apr 7, 2011 at 5:29 AM, Larry Vaden <vaden@texoma.net> wrote:

> >> <learning moment, bitte>
>>
>> What steps are necessary to learn about the reasons for the
>> disconnects/reconnects?
>
> *identifying the reason code for disconnection.
> *enable mac80211 debugs
> *looking at wpa-supplicant logs
> *enabling ath9k debugging
>

Note that in your earlier mail with rc_stats, the majority of packets were
> sent at MCS0 and were not received:
>

>>>> Total packet count::    ideal 34      lookaround 6
>
HT20/LGI TtPMCS0        1.7       28.2       33.3          0(  0)   34
  193

It took 193 attempts to get 34 packets through, around 6 attempts per packet
which is A LOT. The reason the card is disconnecting is almost certainly
that the activity meter at one end or the other is failing -- e.g., the
client is missing a few beacons from the AP and sending a probe that is not
ACKed, or the AP hasn't heard from the client in a large amount of time, and
sends a probe that is not acked. Looking at the dmesg, or wpa_supplicant
logs, or something of that sort, on both sides should tell you the "reason
code" for disassociation. Reason codes are numbers that are mapped to
reasons in the IEEE 802.11-2007 Standard.

Larry, MCS0 is the lowest rate (unless you try 1 Mbps in the 2.4 GHz band).
I really think that if the link can't sustain a reasonable delivery rate at
MCS0 it's simply not going to work.

Are you sure that your nodes have antennas aligned, power levels set to
maximum, etc? No weather or trees in the way? (Did a tree do a springtime
bloom in your LOS, for instance?) [I've never operated long-distance links,
so these are total speculation, btw.]

Dan
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.ath9k.org/pipermail/ath9k-devel/attachments/20110407/96d16751/attachment.html 

[ath9k-devel] why is bitrate of 65.0 MBit/s MCS 7 appropriate for signal strength of -93 dBm?

[Larry Vaden]

On Thu, Apr 7, 2011 at 10:33 AM, Daniel Halperin
<dhalperi@cs.washington.edu> wrote:
> On Thu, Apr 7, 2011 at 5:29 AM, Larry Vaden?<vaden@texoma.net>?wrote:
>
>>> <learning moment, bitte>
>>>
>>> What steps are necessary to learn about the reasons for the
>>> disconnects/reconnects?
>>
>> *identifying the reason code for disconnection.
>> *enable mac80211 debugs
>> *looking at wpa-supplicant logs
>> *enabling ath9k debugging
>>
>
>> Note that in your earlier mail with rc_stats, the majority of packets were
>> sent at MCS0 and were not received:
>>
>> >>>> Total packet count:: ? ?ideal 34 ? ? ?lookaround 6
>
> HT20/LGI TtPMCS0 ? ? ? ?1.7 ? ? ? 28.2 ? ? ? 33.3 ? ? ? ? ?0( ?0)???34
> ? 193
> It took 193 attempts to get 34 packets through, around 6 attempts per packet
> which is A LOT. The reason the card is disconnecting is almost certainly
> that the activity meter at one end or the other is failing -- e.g., the
> client is missing a few beacons from the AP and sending a probe that is not
> ACKed, or the AP hasn't heard from the client in a large amount of time, and
> sends a probe that is not acked. Looking at the dmesg, or wpa_supplicant
> logs, or something of that sort, on both sides should tell you the "reason
> code" for disassociation. Reason codes are numbers that are mapped to
> reasons in the IEEE 802.11-2007 Standard.
> Larry, MCS0 is the lowest rate (unless you try 1 Mbps in the 2.4 GHz band).
> I really think that if the link can't sustain a reasonable delivery rate at
> MCS0 it's simply not going to work.
> Are you sure that your nodes have antennas aligned, power levels set to
> maximum, etc? No weather or trees in the way? (Did a tree do a springtime
> bloom in your LOS, for instance?) [I've never operated long-distance links,
> so these are total speculation, btw.]
> Dan

Good morning, Dan.

THANKS for your reply.

We must reconcile some observations:

. hardware is Ubiquiti Rocket M900 and Loco M900 2 Km apart, estimated
signal strength should be -68 dBm
. AirOS 5.2.1 showed signal strengths in the -70 to -80 dBm range

with, e.g.,

. is the hardware now kaput (see my ? about self-test)
. is there something different about the M900 compared to the M2 and
M5 (from, say, Felix's viewpoint)

Adrian and I did a Google Chat this AM to look at some of the
registers, but, alas, I can't figure out how to cut and paste the
GChat and didn't have the patience to do it stanza by stanza as may be
required.

Here's a few stanzas oriented to learning about what ath9k is telling
the hardware about tx power (as supervised by Adrian):

My apologies in advance if the mechanical transcription of GChat is
misleading or contains errors or omissions (in fact, rather than try
to clean it up, I'll just present the raw segments);  I have included
Adrian on the cc: in case he has more knowledge about how to cut and
paste the GChat transcript.

Adrian:  cat regval
echo "0xA390" > regidx
cat regval
 me:  root at OpenWrt:/sys/kernel/debug/ieee80211/phy0/ath9k# echo
"0xA390" > regidx
root at OpenWrt:/sys/kernel/debug/ieee80211/phy0/ath9k# cat regidx
0x0000a390

Adrian:  cat regval
echo "0xa390" > regidx ; cat regval

me:  root at OpenWrt:/sys/kernel/debug/ieee80211/phy0/ath9k# echo "0xA390" > regidx
root at OpenWrt:/sys/kernel/debug/ieee80211/phy0/ath9k# cat regidx
0x0000a390

root at OpenWrt:/sys/kernel/debug/ieee80211/phy0/ath9k# cat regval
0x1014161c
root at OpenWrt:/sys/kernel/debug/ieee80211/phy0/ath9k# cat regval
0x1014161c

root at OpenWrt:/sys/kernel/debug/ieee80211/phy0/ath9k# echo "0xa390" >
regidx ; cat regval
0x1014161c
root at OpenWrt:/sys/kernel/debug/ieee80211/phy0/ath9k#

 me:  root at OpenWrt:/sys/kernel/debug/ieee80211/phy0/ath9k# echo
"0xa38c" > regidx ; cat regval
0x20242626

I have a second Rocket M900 here 80' from the tower and the intent is
to see what it sees.

Again, THANKS for your reply.

ldv
-- 
Larry Vaden, CoFounder
Internet Texoma, Inc.
Serving Rural Texomaland Since 1995
We Care About Your Connection!

[ath9k-devel] why is bitrate of 65.0 MBit/s MCS 7 appropriate for signal strength of -93 dBm?

[Adrian Chadd]

I just wanted to see if the TX power registers were being programmed
correctly. They look "correct enough" given what I know about how Ubiquiti
quote their power levels.

I really do think the key here is figuring out why Larry is seeing such
crazy low signal strength levels.



adrian
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.ath9k.org/pipermail/ath9k-devel/attachments/20110408/0ecda21f/attachment.htm 

[ath9k-devel] why is bitrate of 65.0 MBit/s MCS 7 appropriate for signal strength of -93 dBm?

[Larry Vaden]

On Thu, Apr 7, 2011 at 7:22 PM, Adrian Chadd <adrian@freebsd.org> wrote:
> I just wanted to see if the TX power registers were being programmed
> correctly. They look "correct enough" given what I know about how Ubiquiti
> quote their power levels.
> I really do think the key here is figuring out why Larry is seeing such
> crazy low signal strength levels.
>
>
> adrian

On the Rocket M900 at 100' from the tower, 5' off the ground with the
AP 80' AGL, we are seeing -86 dBm, granted the antennas on the Rocket
M900 are not per spec, but are AT-5SMA-DB rubber ducks.

IIRC, as a boy in OK, we used to get m?s o menos -85 to -90 dBm if the
string was tight enough between our tin cans :)

ldv

[ath9k-devel] why is bitrate of 65.0 MBit/s MCS 7 appropriate for signal strength of -93 dBm?

[Larry Vaden]

On Thu, Apr 7, 2011 at 7:22 PM, Adrian Chadd <adrian@freebsd.org> wrote:
> I really do think the key here is figuring out why Larry is seeing such
> crazy low signal strength levels.
>
>
> adrian

Does anyone know if these the AR9280 has a self-test?

ldv

[ath9k-devel] why is bitrate of 65.0 MBit/s MCS 7 appropriate for signal strength of -93 dBm?

[Adrian Chadd]

On 8 April 2011 08:35, Larry Vaden <vaden@texoma.net> wrote:

> On Thu, Apr 7, 2011 at 7:22 PM, Adrian Chadd <adrian@freebsd.org> wrote:
> > I really do think the key here is figuring out why Larry is seeing such
> > crazy low signal strength levels.
> >
> >
> > adrian
>
> Does anyone know if these the AR9280 has a self-test?
>

There's a couple of hints I've found in ath9k - eg MAC loop back tests - but
nothing which will test what I think you need to test (ie, the radio.)

Have you replaced the AP unit itself?


Adrian
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.ath9k.org/pipermail/ath9k-devel/attachments/20110408/2a51d6c0/attachment.htm 

[ath9k-devel] why is bitrate of 65.0 MBit/s MCS 7 appropriate for signal strength of -93 dBm?

[Larry Vaden]

On Thu, Apr 7, 2011 at 7:41 PM, Adrian Chadd <adrian@freebsd.org> wrote:
> On 8 April 2011 08:35, Larry Vaden <vaden@texoma.net> wrote:
>>
>> On Thu, Apr 7, 2011 at 7:22 PM, Adrian Chadd <adrian@freebsd.org> wrote:
>> > I really do think the key here is figuring out why Larry is seeing such
>> > crazy low signal strength levels.
>> >
>> >
>> > adrian
>>
>> Does anyone know if these the AR9280 has a self-test?
>
> There's a couple of hints I've found in ath9k - eg MAC loop back tests - but
> nothing which will test what I think you need to test (ie, the radio.)
> Have you replaced the AP unit itself?
>
> Adrian

No, what I think the WISP industry needs is a good self-test :)

I note Felix has made changes, including swatting the 'failed to stop
RX DMA' bug, so I've been compiling an image of r26524 the last 3
hours.

regards/ldv
-- 
Larry Vaden, CoFounder
Internet Texoma, Inc.
Serving Rural Texomaland Since 1995
We Care About Your Connection!

[ath9k-devel] why is bitrate of 65.0 MBit/s MCS 7 appropriate for signal strength of -93 dBm?

[Mohammed Shafi]

On Wed, Apr 6, 2011 at 7:47 PM, Larry Vaden <vaden@texoma.net> wrote:
> On Wed, Apr 6, 2011 at 5:26 AM, Mohammed Shafi <shafi.ath9k@gmail.com> wrote:
>> On Wed, Apr 6, 2011 at 3:28 PM, Larry Vaden <vaden@texoma.net> wrote:
>>> Mindful that we are all ignorant, just about different things, I ask
>>> the ? in the subject line.
>>>
>>> Would that not influence the success ratio and the performance?
>>
>> ath9k rate control is based on PER(packet error rate), but previously
>> it is based on RSSI.
>> But i doubt you are using minstrel rate control, for that please read
>> http://wireless.kernel.org/en/developers/Documentation/mac80211/RateControl/minstrel
>
> With all due respect, I do not think your answer is responsive to the
> question and the data presented.

thanks, I will improve.

>
>>> ? ? ? ?tx packets: ? ? 45
>>> ? ? ? ?tx retries: ? ? 308
>>> ? ? ? ?tx failed: ? ? ?23
>
> Note a FAILURE RATE of 50% and retries ~ 700%.
>
> So, why is 65.0 MBit/s MCS 7 even on the table if rate control is based on PER?
>
> Is it possible that some code other than ath9k is jacking with the
> data rates (read: I am ignorant about what protections OpenWrt might
> have)?
>
> regards/ldv
>
>>> AP (Ubiquiti Rocket M900, running stock r26496):
>>>
>>> root at OpenWrt:~# iw wlan0 station dump
>>> Station 00:15:6d:9c:16:df (on wlan0)
>>> ? ? ? ?inactive time: ?3620 ms
>>> ? ? ? ?rx bytes: ? ? ? 1747
>>> ? ? ? ?rx packets: ? ? 42
>>> ? ? ? ?tx bytes: ? ? ? 3034
>>> ? ? ? ?tx packets: ? ? 45
>>> ? ? ? ?tx retries: ? ? 308
>>> ? ? ? ?tx failed: ? ? ?23
>>> ? ? ? ?signal: ? ? ? ? -93 dBm
>>> ? ? ? ?signal avg: ? ? -105 dBm
>>> ? ? ? ?tx bitrate: ? ? 65.0 MBit/s MCS 7
>>> ? ? ? ?rx bitrate: ? ? 6.5 MBit/s MCS 0
>>>
>>>
>>> root at OpenWrt:~# iw wlan0 station dump
>>> Station 00:15:6d:9c:16:df (on wlan0)
>>> ? ? ? ?inactive time: ?2660 ms
>>> ? ? ? ?rx bytes: ? ? ? 323
>>> ? ? ? ?rx packets: ? ? 11
>>> ? ? ? ?tx bytes: ? ? ? 616
>>> ? ? ? ?tx packets: ? ? 9
>>> ? ? ? ?tx retries: ? ? 49
>>> ? ? ? ?tx failed: ? ? ?3
>>> ? ? ? ?signal: ? ? ? ? -93 dBm
>>> ? ? ? ?signal avg: ? ? -96 dBm
>>> ? ? ? ?tx bitrate: ? ? 6.5 MBit/s MCS 0
>>> ? ? ? ?rx bitrate: ? ? 6.5 MBit/s MCS 0
>>> root at OpenWrt:~#
>>>
>>> CPE (Ubiquiti Loco M900, running stock r26496):
>>>
>>> root at OpenWrt:~# iw wlan0 station dump
>>> Station 00:15:6d:4e:f1:46 (on wlan0)
>>> ? ? ? ?inactive time: ?6950 ms
>>> ? ? ? ?rx bytes: ? ? ? 68637
>>> ? ? ? ?rx packets: ? ? 907
>>> ? ? ? ?tx bytes: ? ? ? 2373
>>> ? ? ? ?tx packets: ? ? 24
>>> ? ? ? ?tx retries: ? ? 88
>>> ? ? ? ?tx failed: ? ? ?3
>>> ? ? ? ?signal: ? ? ? ? -101 dBm
>>> ? ? ? ?signal avg: ? ? -100 dBm
>>> ? ? ? ?tx bitrate: ? ? 6.5 MBit/s MCS 0
>>> ? ? ? ?rx bitrate: ? ? 39.0 MBit/s MCS 4
>>> root at OpenWrt:~#
>>>
>>> best regards/ldv
>>> --
>>> Larry Vaden, CoFounder
>>> Internet Texoma, Inc.
>>> Serving Rural Texomaland Since 1995
>>> We Care About Your Connection!
>>> _______________________________________________
>>> ath9k-devel mailing list
>>> ath9k-devel at lists.ath9k.org
>>> https://lists.ath9k.org/mailman/listinfo/ath9k-devel
>>>
>>
>
>
>
> --
> Larry Vaden, CoFounder
> Internet Texoma, Inc.
> Serving Rural Texomaland Since 1995
> We Care About Your Connection!
> _______________________________________________
> ath9k-devel mailing list
> ath9k-devel at lists.ath9k.org
> https://lists.ath9k.org/mailman/listinfo/ath9k-devel
>

[ath9k-devel] why is bitrate of 65.0 MBit/s MCS 7 appropriate for signal strength of -93 dBm?

[Larry Vaden]

On Wed, Apr 6, 2011 at 9:48 AM, Mohammed Shafi <shafi.ath9k@gmail.com> wrote:
> On Wed, Apr 6, 2011 at 7:47 PM, Larry Vaden <vaden@texoma.net> wrote:
>> On Wed, Apr 6, 2011 at 5:26 AM, Mohammed Shafi <shafi.ath9k@gmail.com> wrote:
>>> On Wed, Apr 6, 2011 at 3:28 PM, Larry Vaden <vaden@texoma.net> wrote:
>>>> Mindful that we are all ignorant, just about different things, I ask
>>>> the ? in the subject line.
>>>>
>>>> Would that not influence the success ratio and the performance?
>>>
>>> ath9k rate control is based on PER(packet error rate), but previously
>>> it is based on RSSI.
>>> But i doubt you are using minstrel rate control, for that please read
>>> http://wireless.kernel.org/en/developers/Documentation/mac80211/RateControl/minstrel
>>
>> With all due respect, I do not think your answer is responsive to the
>> question and the data presented.
>
> thanks, I will improve.

I think your post was important and valuable because it pointed out
that "ath9k rate control is based on PER(packet error rate)" which
seems more sane than EWMA/minstrel IMHO, having operated EWMA/minstrel
for several years.

[ath9k-devel] why is bitrate of 65.0 MBit/s MCS 7 appropriate for signal strength of -93 dBm?

[Mohammed Shafi]

On Wed, Apr 6, 2011 at 8:21 PM, Larry Vaden <vaden@texoma.net> wrote:
> On Wed, Apr 6, 2011 at 9:48 AM, Mohammed Shafi <shafi.ath9k@gmail.com> wrote:
>> On Wed, Apr 6, 2011 at 7:47 PM, Larry Vaden <vaden@texoma.net> wrote:
>>> On Wed, Apr 6, 2011 at 5:26 AM, Mohammed Shafi <shafi.ath9k@gmail.com> wrote:
>>>> On Wed, Apr 6, 2011 at 3:28 PM, Larry Vaden <vaden@texoma.net> wrote:
>>>>> Mindful that we are all ignorant, just about different things, I ask
>>>>> the ? in the subject line.
>>>>>
>>>>> Would that not influence the success ratio and the performance?
>>>>
>>>> ath9k rate control is based on PER(packet error rate), but previously
>>>> it is based on RSSI.
>>>> But i doubt you are using minstrel rate control, for that please read
>>>> http://wireless.kernel.org/en/developers/Documentation/mac80211/RateControl/minstrel
>>>
>>> With all due respect, I do not think your answer is responsive to the
>>> question and the data presented.
>>
>> thanks, I will improve.
>
> I think your post was important and valuable because it pointed out
> that "ath9k rate control is based on PER(packet error rate)" which
> seems more sane than EWMA/minstrel IMHO, having operated EWMA/minstrel
> for several years.
>
thanks! may be a particular rate control is not  always the best in
all conditions/environment.

[ath9k-devel] why is bitrate of 65.0 MBit/s MCS 7 appropriate for signal strength of -93 dBm?

[Larry Vaden]

On Wed, Apr 6, 2011 at 9:59 AM, Mohammed Shafi <shafi.ath9k@gmail.com> wrote:
> On Wed, Apr 6, 2011 at 8:21 PM, Larry Vaden <vaden@texoma.net> wrote:
>> On Wed, Apr 6, 2011 at 9:48 AM, Mohammed Shafi <shafi.ath9k@gmail.com> wrote:
>>> On Wed, Apr 6, 2011 at 7:47 PM, Larry Vaden <vaden@texoma.net> wrote:
>>>> On Wed, Apr 6, 2011 at 5:26 AM, Mohammed Shafi <shafi.ath9k@gmail.com> wrote:
>>>>> On Wed, Apr 6, 2011 at 3:28 PM, Larry Vaden <vaden@texoma.net> wrote:
>>>>>> Mindful that we are all ignorant, just about different things, I ask
>>>>>> the ? in the subject line.
>>>>>>
>>>>>> Would that not influence the success ratio and the performance?
>>>>>
>>>>> ath9k rate control is based on PER(packet error rate), but previously
>>>>> it is based on RSSI.
>>>>> But i doubt you are using minstrel rate control, for that please read
>>>>> http://wireless.kernel.org/en/developers/Documentation/mac80211/RateControl/minstrel
>>>>
>>>> With all due respect, I do not think your answer is responsive to the
>>>> question and the data presented.
>>>
>>> thanks, I will improve.
>>
>> I think your post was important and valuable because it pointed out
>> that "ath9k rate control is based on PER(packet error rate)" which
>> seems more sane than EWMA/minstrel IMHO, having operated EWMA/minstrel
>> for several years.
>>
> thanks! may be a particular rate control is not ?always the best in
> all conditions/environment.

I think the basic problem may be the hardware has gone south or there
is something special about the M900 series because of the down
converter, perhaps.

Now that 24 hours have elapsed without an answer on openwrt-users,
I'll ask here:

Is there a self-test on the AR9280 (or whichever chip is used in the
Ubiquiti M900 series)?

kind regards/ldv
-- 
Larry Vaden, CoFounder
Internet Texoma, Inc.
Serving Rural Texomaland Since 1995
We Care About Your Connection!

[ath9k-devel] why is bitrate of 65.0 MBit/s MCS 7 appropriate for signal strength of -93 dBm?

[Mohammed Shafi]

On Wed, Apr 6, 2011 at 8:48 PM, Larry Vaden <vaden@texoma.net> wrote:
> On Wed, Apr 6, 2011 at 9:59 AM, Mohammed Shafi <shafi.ath9k@gmail.com> wrote:
>> On Wed, Apr 6, 2011 at 8:21 PM, Larry Vaden <vaden@texoma.net> wrote:
>>> On Wed, Apr 6, 2011 at 9:48 AM, Mohammed Shafi <shafi.ath9k@gmail.com> wrote:
>>>> On Wed, Apr 6, 2011 at 7:47 PM, Larry Vaden <vaden@texoma.net> wrote:
>>>>> On Wed, Apr 6, 2011 at 5:26 AM, Mohammed Shafi <shafi.ath9k@gmail.com> wrote:
>>>>>> On Wed, Apr 6, 2011 at 3:28 PM, Larry Vaden <vaden@texoma.net> wrote:
>>>>>>> Mindful that we are all ignorant, just about different things, I ask
>>>>>>> the ? in the subject line.
>>>>>>>
>>>>>>> Would that not influence the success ratio and the performance?
>>>>>>
>>>>>> ath9k rate control is based on PER(packet error rate), but previously
>>>>>> it is based on RSSI.
>>>>>> But i doubt you are using minstrel rate control, for that please read
>>>>>> http://wireless.kernel.org/en/developers/Documentation/mac80211/RateControl/minstrel
>>>>>
>>>>> With all due respect, I do not think your answer is responsive to the
>>>>> question and the data presented.
>>>>
>>>> thanks, I will improve.
>>>
>>> I think your post was important and valuable because it pointed out
>>> that "ath9k rate control is based on PER(packet error rate)" which
>>> seems more sane than EWMA/minstrel IMHO, having operated EWMA/minstrel
>>> for several years.
>>>
>> thanks! may be a particular rate control is not ?always the best in
>> all conditions/environment.
>
> I think the basic problem may be the hardware has gone south or there
> is something special about the M900 series because of the down
> converter, perhaps.
>
> Now that 24 hours have elapsed without an answer on openwrt-users,
> I'll ask here:
>
> Is there a self-test on the AR9280 (or whichever chip is used in the
> Ubiquiti M900 series)?

sorry I don't know about that.

>
> kind regards/ldv
> --
> Larry Vaden, CoFounder
> Internet Texoma, Inc.
> Serving Rural Texomaland Since 1995
> We Care About Your Connection!
>

[ath9k-devel] why is bitrate of 65.0 MBit/s MCS 7 appropriate for signal strength of -93 dBm?

[Daniel Halperin]

On Wed, Apr 6, 2011 at 8:18 AM, Larry Vaden <vaden@texoma.net> wrote:
> I think the basic problem may be the hardware has gone south or there
> is something special about the M900 series because of the down
> converter, perhaps.

I'm confused why you don't think that the problem might be simpler: a
signal strength of -93 dBm is extraordinarily low, and even the most
robust rates may not even work.

Dan

[ath9k-devel] why is bitrate of 65.0 MBit/s MCS 7 appropriate for signal strength of -93 dBm?

[Larry Vaden]

On Wed, Apr 6, 2011 at 11:08 AM, Daniel Halperin
<dhalperi@cs.washington.edu> wrote:
> On Wed, Apr 6, 2011 at 8:18 AM, Larry Vaden <vaden@texoma.net> wrote:
>> I think the basic problem may be the hardware has gone south or there
>> is something special about the M900 series because of the down
>> converter, perhaps.
>
> I'm confused why you don't think that the problem might be simpler: a
> signal strength of -93 dBm is extraordinarily low, and even the most
> robust rates may not even work.

Dan, the units are 2 Km apart, so like Seymour Cray said, when you get
a parity error, you can be sure that one of two things is true.

regards/ldv
-- 
Larry Vaden, CoFounder
Internet Texoma, Inc.
Serving Rural Texomaland Since 1995
We Care About Your Connection!

[ath9k-devel] why is bitrate of 65.0 MBit/s MCS 7 appropriate for signal strength of -93 dBm?

[Larry Vaden]

On Wed, Apr 6, 2011 at 5:26 AM, Mohammed Shafi <shafi.ath9k@gmail.com> wrote:
> On Wed, Apr 6, 2011 at 3:28 PM, Larry Vaden <vaden@texoma.net> wrote:
>> Mindful that we are all ignorant, just about different things, I ask
>> the ? in the subject line.
>>
>> Would that not influence the success ratio and the performance?
>
> ath9k rate control is based on PER(packet error rate), but previously
> it is based on RSSI.
> But i doubt you are using minstrel rate control, for that please read
> http://wireless.kernel.org/en/developers/Documentation/mac80211/RateControl/minstrel

It is my understanding from Changeset 23163 for trunk/package that
minstrel is the default since 09/30/10 17:00:06.

e.g., I don't find an option in menuconfig to change rate control
algos, so I'm pretty what follows reflects the defaults.

root at hq-0900-ap1:~# cat /etc/banner | grep edge
 ATTITUDE ADJUSTMENT (bleeding edge, r26494) ----------
root at hq-0900-ap1:~# dmesg | grep -i rate
ieee80211 phy0: Selected rate control algorithm 'minstrel_ht'
root at hq-0900-ap1:~# date
Thu Apr  7 22:20:02 UTC 2011

Of course, like Dennis Miller, I could be wrong about that.

I'm in the process of trying out Felix's new stuff in 26506, but by
the time I got around to it, svn gave me 26524.

regards/ldv