Forum

Hi all,

I have heard different thoughts on whether your lowest configured data rate should be mandatory or supported. I would love to hear your thoughts from implementations in the real-world or anything else you might have to add.

From what I understand, STAs will rate shift based on transmission errors. These could be caused by too much contention for the wireless medium or by a non-802.11 interferer. So, if you are in an environment with high channel utilization due to beacons and some non-802.11 interference (assuming that it can't be removed entirely), would it be better to set the mandatory rate higher but leave all rates supported to give STAs a better chance of cutting through the interference if it appears that STAs benefit from the lower rate transmissions?

In a different case where you want to reduce channel utilization and there is not significant interference in a given area, if the data rates lower than the lowest mandatory rate are supported, how often will STAs run into problems of not being able to decode beacons, getting out of sync, etc. What other issues does this cause for STAs?

Thanks for your input!!

Getting ready to take the CWNA soon...

David

A little 'bump' on this subject on behalf of my side:

My real world experience so far:

In an environment with high utilization, 'pico cell' AP deployment and some non-802.11 interference i disabled all data rates (for 802.11a/g/n) (80211b is disabled) below 12 Mb and this seems to help a lot. Verified this on the spot with laptop and aircheck analyser. I think because the 'airtime' per STA is much better now.

Any further explanations and advice on this subject highly appreciated!

Best regards, RU

I typically disable all 802.11b data rates. Dont forget if there are b and g client about b clients do not even know the g clients are there so just send packets hence collisions and a poor air environment, let alone a few b clients congesting the network with slow transmissions.

Generally everything under 24mbps is disabled for me.

mcd,

Error rates are not the only determining factor in rate selection. RSSI and other factors could also be used.

There is often a good corelation between rate adaptation and a devices roaming algorithm. The specific choices are usually proprietary although, as time goes on, players seems to be learning from each other to some extent.

Simplistic roaming and rate adaptation algorithms based on single factor measurements may work in the lab, but usually fail to work that well in the real world.

Be caerful with turning of all low speeds. This will create cells with very step drop drop in coverage areas at the edges of a cell. This causes problems when for exmple VoFi clients turns around a corner in a hallway and suddenly have a low signal strength and no usuable speed to connect to, even if the signal strength could have been good enough for a slower speed. Add to this the fact thet a handset may be shielded by the body depending on the walking direction of the caller you may experience delays in finding a roaming alternative.
If you have a roaming treshold of -67 RSSI and you can not connect to for example 18 Mbps caused by interference then you have problems.

As you all may know laptop users may not consider this a problems since most of their applications uses TCP instead of UDP and thus can tolerant huge amount of traffic delays. This may though change when we goet more and more handheld devices depending on WiFi access.

mericson

Great point certain vendors now recomend that aps be placed at transition points specifically for that reason to assist voice roaming. I actually only found this out yesterday on a conf call because this traditional survey is seeing some issues where the cells are fine but the roaming algorith isnt quite intelligent enough.