Prism2 Cards: Maximum Distance / Ack Timeout?

George Heeres george at msworldnet.com
Thu Mar 9 10:12:26 EST 2006


I've got an Access Point running on a MikroTik RouterBoard 532 with a
200mW Prism2 card (NL-2511MP) on a 12 dBi Omni using MikroTik 2.9.10
(was running 2.9.13 but had to downgrade due to OSPF problems) powered
by 48V POE. From what I've been able to determine by looking at the
internal file system, the MikroTik appears to be using the linux-wlan
drivers (although I don't have confirmation of this, nor do I know what
version...or if customized).

I've got two clients, which are running on WRAP.2C boards, with a custom
built Linux operating system (2.6.11.7 kernel) based on uClibc using the
HostAP drivers (CVS checkout from sometime in early 2005). These clients
are running the exact same card as the Access Point (200mW Prism2 card,
NL-2511MP) with a 14 dBi RooTenna powered by 12V POE.

The distances for these two clients from the access point is: 4.79 mi
and 6.51 mi respectively.

Each of these clients is seeing a signal strength of roughly -80/-84
with a -92/-94 noise floor (roughly same signal strengths from both
tower / client). I'm seeing ping times all over the place with dropped
and duplicate packets:

64 bytes from 10.1.18.251: icmp_seq=1 ttl=62 time=6.40 ms
64 bytes from 10.1.18.251: icmp_seq=2 ttl=62 time=60.6 ms
64 bytes from 10.1.18.251: icmp_seq=3 ttl=62 time=55.8 ms
64 bytes from 10.1.18.251: icmp_seq=5 ttl=62 time=106 ms
64 bytes from 10.1.18.251: icmp_seq=5 ttl=62 time=330 ms (DUP!)
64 bytes from 10.1.18.251: icmp_seq=7 ttl=62 time=5.72 ms
64 bytes from 10.1.18.251: icmp_seq=8 ttl=62 time=52.9 ms
64 bytes from 10.1.18.251: icmp_seq=9 ttl=62 time=67.0 ms
64 bytes from 10.1.18.251: icmp_seq=10 ttl=62 time=73.3 ms
64 bytes from 10.1.18.251: icmp_seq=12 ttl=62 time=6.16 ms
64 bytes from 10.1.18.251: icmp_seq=13 ttl=62 time=154 ms
64 bytes from 10.1.18.251: icmp_seq=13 ttl=62 time=217 ms (DUP!)
64 bytes from 10.1.18.251: icmp_seq=14 ttl=62 time=45.9 ms
64 bytes from 10.1.18.251: icmp_seq=15 ttl=62 time=233 ms
64 bytes from 10.1.18.251: icmp_seq=16 ttl=62 time=43.6 ms
64 bytes from 10.1.18.251: icmp_seq=17 ttl=62 time=73.2 ms
64 bytes from 10.1.18.251: icmp_seq=18 ttl=62 time=5.93 ms
64 bytes from 10.1.18.251: icmp_seq=19 ttl=62 time=60.4 ms
64 bytes from 10.1.18.251: icmp_seq=20 ttl=62 time=7.10 ms
64 bytes from 10.1.18.251: icmp_seq=21 ttl=62 time=14.2 ms
64 bytes from 10.1.18.251: icmp_seq=22 ttl=62 time=5.86 ms
64 bytes from 10.1.18.251: icmp_seq=23 ttl=62 time=207 ms
64 bytes from 10.1.18.251: icmp_seq=24 ttl=62 time=84.8 ms

Browsing around online, I found mention from someone regarding ACK
timeouts using the PRISM2 based cards beyond 8 km (4.9 mi). Additional
reading has confirmed that the ACK timeout on the PRISM2 based cards is
fixed. The nature of the problem (duplicate / dropped packets, etc.)
seems to confirm this as the problem.

I'm trying to determine if the ACK timeout is the cause of the problem
or if it's something else before I start replacing cards. Here is what
I'm thinking as potential reasons for the problem:

1.) ACK timeout. Is there anything that can be done or I just need to
use different cards (ie. Atheros based with dynamic / adjustable ACK)
2.) Signal just not strong enough? What should I be shooting for?
3.) Incompatibility between linux-wlan and hostap drivers? (Don't think
so...)

Please help shed some light on this for me. I'm at the limit of my
knowledge regarding wireless. If you need additional information, please
let me know what you need and I'll do my best to make sure you get it.

Thanks
George



More information about the HostAP mailing list