[olsr-dev] Re: [OLSR-users] Differing metrics for route decision

Sven-Ola Tuecke (spam-protected)
Tue Feb 28 12:34:36 CET 2006


yes - but we still need a couple of days experience to dig the implications. 
This night my two testbed WRTs broke out and caused a major BSSID split 
here. For whatever reason they adoped the TSF of 500.000 years for a couple 
of hours. This morning they both went through the TSF overflow and now 
situation is normal again (The mesh itself uses another BSSID currently):

(spam-protected):~# freifunk-setbssid -t eth1
00000002:7c4e6799 (0 years, 0 days, 02:57:55)
(spam-protected):~# wl rateset
rateset : [ 1(b) 2(b) 5.5(b) 6 9 11(b) 12 18 24 36 48 54 ]
(spam-protected):~# wl bssid
bssid is 02:CA:FF:EE:BA:BE

Some things to dig:

The rateset is normally taken from the oldest node. Because this is most of 
the time a rotten B-mode-only PCMCIA we do not have g-mode in the mesh. To 
my observation it is possible to have g- and b- mixed in the mesh, e.g. 
talking g to one side and b to the other. Basic (b) rates in the rateset is 
used for beacons (and bcasts?). "wl gmode=4" will have additional basic slow 
rates with g-mode (e.g. the 6 mbit rate is G -> declared basic with 
gmode=4). Should that be used everywhere? There is a multicast-rateset also 
with Broadcom, don't know if this will influence the rate-selection for 
broadcasts. olsrd broadcasts will need to send b-mode to address all 
stations. I'am not too satisfied with the auto-rateset determined with 
received beacons/packets (manual tweaks for a specific connection usually is 
better) but the auto-adaption to current air condition should be preserved. 
For this, I may need to tweak the selective-beacon-deafness currently used 
for the fixed-BSSID hack. Have read that OFDM is superior to b-mode (with 
equal speeds, e.g. 5b vs. 6g) in terms of reflections / polarization - so 
should g-mode be preferred really for long-links?

LG Sven-Ola

"Aaron Kaplan" <(spam-protected)> schrieb im Newsbeitrag 
> that is _wonderful_ news !
> wow
> On Feb 27, 2006, at 2:34 PM, Sven-Ola Tuecke wrote:
>> Other remark: This weekend I was able to solve the BSSID-split-in- Ad-Hoc 
>> thing for Broadcom Wifi. Have two WRTs talking to each other  in my 
>> bedroom on our olsr.freifunk.net SSID, but they don't care  about other 
>> becons flying around. Means: No more unpredictable  channel switches, no 
>> BSSID wars, no fallback-to-b-mode-only, no  hacking with TSF=500.000 
>> years etc. To be evaluated next wednesday  in the c-base Wifi-hell where 
>> a couple of PCMCIA card owners  normally bring the local area down :)
> _______________________________________________
> olsr-dev mailing list
> (spam-protected)
> https://www.olsr.org/mailman/listinfo/olsr-dev 

More information about the Olsr-dev mailing list