[orbit-user] Detecting Wireless Collisions

[Mesut Ali Ergin]

Rick,

If you are in a controlled environment with little or no interference 
(i.e., a clean 802.11 channel), and if you operate at a PHY rate that 
assures enough SNR (with a margin for safety) at your receivers, 
majority of the losses will be collisions. For example, if you pick a 
non-occupied 802.11a channel on the grid and use 6Mbps as your link 
rate, almost all retransmissions will be due to collisions (minus driver 
bugs and bad hardware). See below paper for a more detailed discussion 
on another way to tell.

Experimental Assessment of 802.11 MAC Layer Channel Estimators.
D. Giustiniano, D. Malone, D. J. Leith and K. Papagiannaki.
IEEE Communications Letters, 2007.
http://www.hamilton.ie/net/channelestimation2007.pdf

Best,

-- 
Mesut Ali Ergin
ergin at winlab.rutgers.edu

Rutgers University, WINLAB,
Technology Centre of New Jersey,
671 Rt. 1 South, North Brunswick,
New Jersey, 08902-3390, USA

Phone: 862-368-6620
Fax:   732-932-6882


Rick Correa wrote:
> Does anyone know a good way to detect wireless collisions?
> 
> We tried looking at collisions as reported by ifconfig:
> ath0      Link encap:Ethernet  HWaddr 06:60:b3:ac:2b:fe
>                   inet addr:192.168.7.12 <http://192.168.7.12>  
> Bcast:192.168.255.255 <http://192.168.255.255>  Mask:255.255.0.0 
> <http://255.255.0.0>
>                   inet6 addr: fe80::460:b3ff:feac:2bfe/64 Scope:Link
>                   UP BROADCAST RUNNING MULTICAST  MTU:1500  Metric:1
>                   RX packets:1016144 errors:0 dropped:0 overruns:0 frame:0
>                   TX packets:22216 errors:0 dropped:0 overruns:0 carrier:0
>                   collisions:0 txqueuelen:0
>                   RX bytes:182508688 (174.0 MiB)  TX bytes:4129076 (3.9 MiB)
> 
> 
> But it always returns 0.  If we get a packet capture, we see some 
> packets have the Retry bit set to 1 in the 802.11 frame.  Does this 
> imply it is a retransmission caused by a collision?  Is there any other 
> way to get collision stats?  We looked at iwlist and iwconfig but we 
> couldn't find a useful utility.  Perhaps something in /proc?
> 
> 
> ------------------------------------------------------------------------
> 
> _______________________________________________
> orbit-user mailing list
> orbit-user at orbit-lab.org
> http://orbit-lab.org/cgi-bin/mailman/listinfo/orbit-user