Hallo all!
In Orinoco APs, there is parameter "Distance between APs" (Large, Medium, Small,
Minicell, Microcell). This value influences Modem Thresholds
(EnergyDetectThreshlod, CarierDetectThreshold and DeferThreshold) and Roaming
Thresholds (from man page of last linux driver)
In Windows driver, there is parameter SystemScale tunable via regedit.
So, if there is big interference, after some lowering of "sensitivity" is still
possible to use classic 802.11b mode at small distances or for not so long
distances with directional antennas.
802.11b uses CSMA/CA - so only one station is able to transmit and rest of
stations must wait (while value of backoff timer will be zero)
Now questions:
1) Turbocell: MAC level is still based at CSMA/CA?
2) If yes, is there any advantages of Turbocell with access to backoff timer
(possibility to start with lower values), so Turbocell packets will have better
priority than classic 802.11b packets. Spectralink VOIP supported in classic
802.11b APs is based at zero value for backoff timer (i think).
3) Sensitivity - is there any options to change Turbocell SystemScale
(sensitivity) or it is fully automatic = Since Turbocell kernel V4 my MRTG
Signal level graphs are unstable: may be "sensitivity" issue?
Because, if there is no sensitivity tuning, turbocell can not compete with
classic 802.11b mode in environment with high interference.
4) How about two or three ISP Base stations with two 180 degree (or three 120
degree) antennas at same channel placed at same antenna mast. Will this work OK?
Throughput is for me not important. Important is quality at low speeds and
advantage against classic 802.11b. Will "polling" work OK, if there will be
several base stations at same place?
I tried simultaneous running of Turbocell Base and 802.11b AP mode at same place
with two 360 degree antennas and noticed no problems. Interference between
802.11b and Turbocell at same antenna mast was eliminated with CSMA/CA (i
think). Of course there is "hidden node problem", but this will be allways,
because it is mostly caused by devices from competitors.
5) Turbocell is using "polling", but Satellite must still wait, while medium is
free? In many cases is possible, that even if packet will be transmitted while
medium is not free, it could be successfully received.
6) Backoff timer: i still do not know how this mechanism is working, if there
are 2 wireless devices (not so far placed) at adjacent channels and what could
be even worse are from different vendors, so first device is able to hear
second, while second can not hear first.
7) Or even worse, there are diversity antennas used... PRISM cards are often
with 2 antenna connectors, some Karlnet support for selection which connector is
for output and which for input will be nice.
May be some answers are secret = OK. But i think, that tunable access to backoff
timer = priority, and tunable "sensitivity" could be solution how to tune
Turbocell to be more aggresive.
In this message i have only one goal = better Turbocell software.
Regards
Peter
_______________________________________________
Karlnet mailing list
Karlnet@WISPNotes.com
http://lists.wispnotes.com/mailman/listinfo/karlnet
|