Measuring Wireless Performance (?)

[Sam Leffler]

Ronald F. Guilmette wrote:
> As I just mentioned in my immediately preceeding post, I'm a total
> neophite when it come to wirless networking, so I need to ask a
> rather basic question.
>
> In preparation for installing my first ever wireless network, I read
> up on the subject awhile first, and I found several people who had
> commented (in various places) that they had bought "upgrade" antennas
> for their wirless cards, and that this helped them, either to make
> connections where they otherwise couldn't, or else with throughput/
> performance of their wireless link.
>
> Being totally new to this stuff, I have no idea if I would benefit
> from a better antenna for my wireless card or not, so I gotta ask:
> How can I tell?
>
> Are there some tools available for FreeBSD that would tell me about
> stuff like:
>
>     Received Signal Strength Indicator (RSSI)
>     Failed Frame Transmission Count, etc.
>
> and/or performance of the wirless link generally?
>
> Humm... OK.  I just found this, but it seems to be a Windoze-only thing:
>
>    http://sysnet.ucsd.edu/pawn/wrapi/
>
> :-)
> _______________________________________________
>   
ifconfig ral0 list sta shows rssi + noise floor if the driver provides 
it.  As to performance you need to distinguish between tx+rx operation.  
The ral driver in REELNG_7 uses a tx rate control algorithm that is 
mismatched for the device capabilities.  As a result tx performance is 
suboptimal.  Using unidirectional UDP tests like netperf -t UDP_STREAM 
in each direction is useful to characterize operation + performance.

If your problem is you are simply too far from your ap then boosting 
gain with an outboard antenna can help but rx sensitivity is just as 
important (in many cases more important) than tx power so whatever you 
do must consider this.

In the end you're likely better off just springing for a better wireless 
card.

    Sam