> I am afraid I must fall on my sword, so to speak, in this case. It
> appears that bob is 100 percent correct about the compatibility issue.
> I slap myself for not realizing that for a radio push 5 GHZ
> it needs the extra voltage.
Not only that. CardBus is the 32-bit successor of 16-bit PCMCIA. That is,
these a and g cards use it for performance reasons. Same as with the
transition from 10 Mbps (16-bit) ISA cards to 100 Mbps (32-bit) PCI cards
...
> I am still trying to get my backbone shots off of b as
> the noise floor around my neck of the woods is incredible, something
> around -75 dbm!! I have yet to decide if this is directly
> related to my own amount of presence in the area, or other 2.4GHZ
equipment
Maybe that the US has much more 2.4 GHz usage than Germany, but whenever we
have noise that high, it's usually one of two things:
a) our own use of the spectrum: this is sometimes from near/same-channels we
use a few kilometers(!) away. It all depends on antenna gain, direction and
the amount of traffic - you will see a pattern that correlates to peaks in
traffic in other segments (even if it is from competitors).
b) TV/VCR equipment: Lately, we have had instances of someone using such
stuff inhouse exactly in the focus of a narrow beam antenna or very close to
the CPE antenna. The pattern is usually: noise appears in the evening and
disappears later in the night (except if they never turn it off again), but
it's never correlated to traffic peaks.
To free 2.4 frequencies and to boost the backbone, which is typically made
from PTP links, "normal" 802.11a/g equipment is usually much better than a
polling (Karlnet, Proxim) solution because RTT is a stable 2-4 ms and
performance is better. There is really no need to run TC on PTP links
(except because the old Orinoco implemented bridging this way). Modern WDS
featured bridges will do better if you stick to PTP.
While this sounds like "just pick any 11a AP with WDS-bridging", we've found
that there are only a few devices which support something close enough to
"management". The tendency being "hide signal/noise readings and the
customer will be happy", there is only one device I've seen which has a
"wireless link test" feature with signal, noise and SNR readings/measurement
like the Orinoco / Karlnet devices, combined with SNMP, WEB and TELNET
configuration, PoE, etc.: The LANCOM L54ag from Germany. There may be others
in the US and it's not my intention to have you buy from Germany ;-)
although this unit would support US regulatory requirements for 802.11a/g as
well and it supports 802.11h for the 5.475-5.725 GHz range which is being
considered by the FCC (for 11 more channels).
We have done measurements: 802.11a throughput is 24-28 Mbps on the LANCOM
L54ag if you manage to keep an RSL better than -66 dBm and a SNR better than
20 dB. With that many channels to choose from, you can even run Atheros 108
Mbps mode and get 48 Mbps net throughput (also measured) but you need to
have somewhat better SNR for 108 Mbps mode.
Just in case some readers forgot what I said upfront: This is only for PTP
links; doing PMP with the LANCOM will give you hidden node problems ... that
is still the domain of Karlnet, if Karlnet doesn't miss out while Proxim is
already shipping its MP.11a which uses the WORP polling mode on 5.x GHz ...
--
true global communications GmbH
Thomas Giger
In der Au 27, 61440 Oberursel, Germany
fon +49.6171.6381-0, fax +49.6171.6381-19
www.tgnet.de || www.megaspeed-internet.de
|