Testing strategies for 802.11 networks

Hi,

I require a testing strategy for an 802.11 wireless network. The
objective is to provide some sort of QoS. If you know of any existing
testing strategy/framework or something suitable at least, please let
me know. Suggestions appreciated.


Regards
Patrick

patrick_larry_mcgrath@yahoo.ie (Patrick) wrote:

>I require a testing strategy for an 802.11 wireless network. The
>objective is to provide some sort of QoS. If you know of any existing
>testing strategy/framework or something suitable at least, please let
>me know. Suggestions appreciated.

It is difficult to image any strategy for obtaining a meaningful QoS.
Either you make optimistic assumptions and get very good data rates,
or pessimistic assumptions and get very bad data rates.

Unlike a wired network, you can't assume that all the signals are
under your control (i.e. generated by your equipment), and a wireless
network has the capability for very severe degradation when there is
excessive traffic and/or poor signal strength and/or interference.

Our customers do use WLANs in time-sensitive applications, but the
error-handling strategies in the higher protocol layers do need some
thought.

Jeremy Bentham
Iosoft Ltd.

Patrick wrote:
> I require a testing strategy for an 802.11 wireless network. The
> objective is to provide some sort of QoS. If you know of any existing
> testing strategy/framework or something suitable at least, please let
> me know. Suggestions appreciated.

Your high-level strategy depends on what you mean by quality of service. 
  You can measure bandwidth, signal margin, or latency, to mention a few 
criteria.  Are your devices mobile?  Is it important that QoS (whatever 
you define that to be) remain constant with distance?  Are the devices 
constrained to a given geographical area?  Does that area have 
sigificant natural and artificial features that will affect signal paths?

For geographical analysis, for example, to determine good locations for 
cell phone towers, there are a few commercial software packages that you 
can license.  These tools use various propagation models and can be 
handy for doing what-if analyses on a fixed geographical area, assuming 
you have decent maps.

But even with good software models, you'll end up going out to the field 
to do some measurements to do on-site validation.

An easy way to control the environment in the lab is to connect the 
devices together via a transmission line.  This allows you to easily 
insert calibrated attenuations and/or interference.

Kelly