[Wolves] Wireless network access

leo sandhu leosandhu at gmail.com
Wed Oct 4 02:35:58 BST 2006


Is this thread still active and relevant?   I apologise if not.

You have a lot of variables.  Remember that if using the 2.4Ghz airwave
spectrum, there are 2 main standards, the 8-2.11b and 802.11g.   You will
find that most domestic 2.4ghz access points (AP) are now .11g as this
supposedly delivers speeds of 54mbps.
On the other hand, .11b is limited to 11mbps but is actually stable (big
advantage really or else the network will be very happy until end-users
actually attempt use).

IMHO things like big walls and thick carpets will be useful if you want to
provide a more reliable indoor network.   Especially if you want it to
actually perform well in every room that you cover.

My suggestion would be that you place one AP in each room, something like a
Soekris machine with a mini-pci wireless card and loaded with something like
M0n0wall on a read-only cf card will do the trick very nicely.

The next 2 important options are your antennae (honest diy homebuilds ARE
cool) and the allocation of channels to the individual APs.  If you want to
fill an entire building with solid wifi at reasonable performance, many low
power antennae are going to serve you better than one whopper that chokes
after the first dozen or so users jump aboard....

It is useful to plan your position of APs based upon the results of a sweep
with something like Wi-Spy, which is a spectrum analyser that although
packaged for M$ works with several Linux forensic utilities.

PS, as the 802.11a - 5Ghz range has already been mentioned, it would be
possible to use something like M0n0wall with more than one mini-pci so one
.11b and one .11a could occur with shared policies.... Unfortunately I
haven't seen any spectrum analysers for the 5ghz range so planning maybe
more problematic - that said, not many people actually have .11a equipment
as the performance is supposedly lower than the .11b

If this helps any, I'd be happy to know ;)   For the record, you guys all
missed a great week in Leiden this August when the Wireless OpenAir Festival
took place.   People who attended that event are now developing wifi
networks across Tanzania and using the hardware and router knowledge that
was exchanged to expand wifi as a credible solution for building open
networks.

leo.

On 9/8/06, Darran Rimron <darran at xalior.com> wrote:
>
> On Fri, 2006-09-08 at 11:05 +0100, Mark Ellse wrote:
> > We've got a Buffalo domestic wireless router that provides reasonably
> > good wireless internet access around one common room. But our walls are
> > thick and the range is poor. What should I do, go for a number of
> > similar routers scattered around the place, or something more industrial
> > and powerful? Does anyone have any experience?
>
> IMO (YMMV - these are from _my_ _personal_ playings, and hardly
> scientific):
>
> * 802.11a rocks (aka mmm5Ghz! *drool*) for this sort of deployment, but
> cheap, easily available kit is rarer than rocking horse droppings and
> not all client chip-sets support this.
> * 802.11b does better at long (open) distances if you don't mind the
> speed loss.
> * 802.11g if noticeably worse if you have "b/g compatibility mode"
> enabled, killing "backward compatibility" gives results far superior to
> 802.11b.
>
> If you get more "domestic wireless routers" there's a good chance that
> they won't do "passing off" from one access point to another, I've had
> problems with this before on cheap kit. I ended up buying WRT54Gs and
> throwing White Russian on them.... However multiple access points (on
> different channels) can often deal better with "Microwave
> Pollution" (from DECT phones or good, old-fashioned Microwave ovens) in
> the 2.4Ghz frequency, especially the lower channel numbers....
>
> If I was in your position I'd see if your current access point supports
> things like ad-hoc passing of clients between base-stations, etc. This
> is often firmware dependant at the _client_ end, and thus sucks the
> donkey dong :( the easiest way to test is to nick an access point from a
> mate and intentionally switch your client between the two.
>
> In short - multiple access points is going to give you a more reliable
> and "hardened" network for general day-to-day use, if you happen to have
> a wifi card that is clever enough to go, "Oh, look, multiple access
> points on the same SSID, with the same crypt details but on different
> channels, I'll pick the strongest...." - HOWEVER some (the
> cheaper/commercial) APs cache your MAC address on the AP itself and when
> you roam from one access point to the second you quite often you can't
> get access to "the net" until the ARP cache on the first AP has expired.
> (this is what was happening at the LRL06 hotel :( )
>
> There's no "easy" way to check this, without nicking an access point
> from "elsewhere" - for example, I found my WRT54G did exactly this until
> I reblew the firmware with WhiteRussian :)
>
> Of course, in some cases, it doesn't matter how beefy your transmitter
> is, steel steam pipes, air conditioning ducts or a high density of
> organic material (human bodies are a great radio dampener) all are going
> to beat the living daylights out of your signal strength anyway. (At
> LRL06 we had 80odd percent signal from one side of the room to the
> other, until the punters arrived, and then we couldn't get 4 ping
> packets in a row across it....)
>
> I would:
> * Evaluate your current hardware: does it cache MAC/ARP; and thus
>    making it useless for localised roaming between same SSID
> * Evaluate your environment: do you have objects that are going to
>    kill wifi no matter how strong it is; Metal and Meat are the big
>    no-no's here.
>
> If you find option 1 is the case, an additional AP will help, but will
> introduce some rather painful/annoying network issues, however, no
> matter how good your antenna if you're finding yourself in option 2,
> you're gonna need multiple transmitters....
>
> While no proper answer, I hope this helps someone.... :)
>
> -Dx
>
>
>
>
> _______________________________________________
> Wolves LUG mailing list
> Homepage: http://www.wolveslug.org.uk/
> Mailing list: Wolves at mailman.lug.org.uk
> Mailing list home: https://mailman.lug.org.uk/mailman/listinfo/wolves
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mailman.lug.org.uk/pipermail/wolves/attachments/20061004/0e520047/attachment.html


More information about the Wolves mailing list