On 5 Oct 2004 18:43:32 -0700,
(E-Mail Removed) (Jacobs
Scooter) wrote:
>I was wanting to know how bandwidth is distributed on 802.11g wireless
>access points amongst simultaneous users?
In theory, the bandwidth is distributed equally to each user. 802.11g
will yield about 30Mbits/sec of thruput. Each users gets an equal
part of the pie.
However, this assumes identical client radios, identical signal
strength, identical traffic patterns, and identical packet sizes. It
also assumes that there are no hidden nodes, which will cause
collisions, that there are no high traffic related anomalies in the AP
firmware, and that the overhead is minimal. The usual method of
mitigating collisions from hidden nodes is to enable CTS/RTS flow
control on the wireless side. This is normally disabled and
un-necessary for typical small WLAN's. Turn on flow control and your
bandwidth gets can easily get cut in half. Leave it off and
collisions will do the same thing, possibly worse.
>Am I incorrect in assuming
>that if I need for each one of my 2000 simultaneous users to have
>reliable duplex speeds of 750Kbps I would need a 2000 x 750Kbps = 1500
>Mbps? Or if each 54Mbps 802.11g access point supports 250 SU's that
>each user has only 216Kbps available to each of them?
How about I don't answer the question and ask you a few in return?
What application sucks 700Kbit/sec that 2000 users would want to use
at a fair? Are all of them going to use this bandwidth at the same
time? If not, what percentage of the time? In what direction
(download, upload, or both)? (Know anything about queueing theory?)
Well, those are rhetorical questions, but they should give you a clue
what to ask. My guess(tm) is that you'll have no more than 10%
loading and can easily live with 128-256Kbits/sec per user per access
point when active. You'll be stuck with using 802.11b at 5.5Mbits/sec
because of the increased range an penetration of the lower speeds.
That gives you about 200 users per access point. If you put 3 radios
on each poletop, that will give you 600 users per hot spot. If you
try to deliver anything more or faster, you will have problems. The
aggregate bandwidth at each poletop will be about 6-10 Mbit/sec which
can easily be handled by just about any kind of transparent bridge
(that can handle 600 MAC addresses). Your backhaul to the ISP will be
6-10Mbits/sec times the number of hot spots. Note that this is only a
guess based upon what little information you've supplied.
>Also if I have a
>3Mpbs T3 line hooked into a 802.11g 54Mbps wireless access point can I
>really expect speeds of 54Mbps or am I limited to the 3Mbps to its
>internet backhaul. I don't see how a wireless access point can speed
>up data rates but perhaps that's one of the secrets of the radio
>technology that's involved.
Ahem. If you backhaul is also on 2.4GHz, you're more limited than
that. Since each packet will appear in the same airspace twice, your
aggregate bandwidth just got cut in half. Add the store-n-forward
overhead of your mesh network, which sends additional identical
packets in the same airspace, and you will soon have no bandwidth
left.
>Just looking for some clarification. Thank you all for your input.
>Jacob
--
# Jeff Liebermann 150 Felker St #D Santa Cruz CA 95060
# 831.336.2558 voice
http://www.LearnByDestroying.com
#
(E-Mail Removed)
# 831.421.6491 digital_pager
(E-Mail Removed) AE6KS