[Olsr-dev] mac80211 / link-bandwith from userland for costcalculation / plugin

Teco Boot (spam-protected)
Wed Sep 15 12:27:02 CEST 2010


This clearly shows the difficulties using link metrics.
Neighbor 10.10.54.1 has 
 - high loss
 - low signal strength (I think converted RSSI, which is *not* signal strength)
 - a moderate 5.5mbit data rate.
The data rate is not what could be expected. Probably there was no unicast traffic, 
and rate selection took place some time ago, with other conditions. Unicast would 
take alternate path, as ETX=11.76 would be higher than 2-hop with cost=~1.
A unicast neighbor data rate probing mechanism would update the rate.

Your info shows asymmetry in loss. Does this correspond with RSSI on other ends?

Madwifi provides:
olsr-rt02:~# wlanconfig ath0 list  
ADDR               AID CHAN RATE RSSI  DBM  IDLE  TXSEQ  TXFRAG  RXSEQ  RXFRAG CAPS ACAPS ERP    STATE     MODE
00:0c:42:68:6d:bf    0    1  36M   16  -79     0   2033      12    585       0              0        1   Normal
00:0c:42:68:6d:b5    0    1  36M   17  -78     0    338       1   3537       0              0        3   Normal
00:0c:42:68:6d:ba    0    1  36M   53  -42     0    340       7   3211       0              0        3   Normal
00:0c:42:39:e8:2e    0    1  36M   26  -69     0    340      11   2474       0              0        3   Normal
00:0c:42:39:e8:28    0    1  36M   31  -64     0    350      12   1921       0              0        3   Normal
00:0c:42:68:6d:ae    0    1  36M   21  -74     0    363       7   3768       0              0        3   Normal
00:0c:42:68:6d:cd    0    1  36M   36  -59     0    342       8   2211       0              0        3   Normal
00:0c:42:68:6d:c0    0    1  36M   52  -43     0    340       8    388       0              0        3   Normal
00:0c:42:68:6d:cc    0    1  36M   22  -73     0    347       6    343       0              0        3   Normal
The presented rate 36M is false. With older version, value was correct.
But had other problems...

Teco


Op 14 sep 2010, om 09:52 heeft Bastian Bittorf het volgende geschreven:

>> Outgoing datarate and incoming RSSI is always "per neighbor", at
>> least if you 
>> query them with the "iw station dump" command (or a similar nl80211
>> code)l.
>> 
> 
> here some interesting output from our debugging tool (with rssi/mbit):
> 
> 
> (spam-protected):~# ./neigh.sh 
> 10.10.60.1      10.10.2.1       0.00    1.000   0.788   1.269    de:5d:4c:a4:f0:ac -65dBm 1.0mbit 
> 10.10.60.33     10.10.2.33      0.00    1.000   1.000   1.000    ethernet  
> 10.10.60.33     10.1.0.8        0.00    1.000   1.000   1.000    ethernet  ***gateway***
> 10.10.60.1      10.10.18.1      0.00    0.862   0.650   1.780    9a:0c:6d:fe:4b:a6 -90dBm 1.0mbit 
> 10.10.60.1      10.10.39.1      0.00    0.788   0.862   1.470    9a:0c:6d:fe:3c:2a -88dBm 1.0mbit 
> 10.10.60.1      10.10.61.1      0.00    0.823   0.596   2.037    06:27:19:ec:cf:46 -79dBm 18.0mbit 
> 10.10.60.1      10.10.17.1      0.00    0.732   0.000   INFINITE 06:27:19:ec:cf:46  
> 10.10.60.1      10.10.54.1      0.00    0.223   0.380   11.760   9a:0c:6d:fe:5d:12 -92dBm 5.5mbit 
> 
> 
> -- 
> Olsr-dev mailing list
> (spam-protected)
> http://lists.olsr.org/mailman/listinfo/olsr-dev





More information about the Olsr-dev mailing list