[MLB-WIRELESS] Channel management

[Simon Butcher]

Hello Mr. Bunny :)

> SB> 1. Channels 1 through 10 have a maximum power output of 4 watts 
> SB> EIRP, however channels 11 through 13 have a maximum of 200 
> SB> milliwatts EIRP. Obviously, a link which only spans a 
> short distance 
> SB> should use the latter segment.
> 
> 1 to 9, 10 to 13 (this is under review by the ACA at present 
> to make it uniform of 4W across the entire band)

Ahh, I must have gotten my wires crossed there then, sorry. I was given
a manual for an old no-name brand 802.11 card (not 802.11b), and it says
channel 11 is 2462 MHz (the ACA's boundary between the two EIRP limits
point is 2463 MHz, right?).

The manual mentions no brand name for the card, and the information
could be entirely wrong - Sorry :(

It would be nice if the ACA do make it 4 watts across the board - I've
met a lot of network admins who have set up WLAN's on channels above 10,
and weren't aware of the 200 mW limit..

> SB> 2. The more networks on the same channel, the more packet 
> dropping 
> SB> (from mismatching SSID's) must be done. This is 
> counter-productive. 
> SB> Unfortunately outside sources will also undoubtedly 
> effect this too, 
> SB> such as commercial hot-spots and so forth.
> 
> Erm I don't think this quite what happens... any number of 
> other things that can generate noise without being different 
> SSID's... I don't think you have a hope in hell of even being 
> practical in this respect...

I wasn't talking about noise. If you have two networks operating in the
same area, but with different SSIDs, then a card still has to receive
the header, see that it's got another SSID, and ignore the radio waves
for the time period denoted in the header - this can reduce the
through-put enormously (and sometimes introduce some synchronisation
problems).

Of course there are sources of noise from other things which can kill
through-put too, but having two visible networks with differing SSID's
on the same channel is an obvious bandwidth killer.

> SB> 3. Using channels sequentially in an area is also 
> counter-productive 
> SB> due to noise generated by channel cross-talk. I believe it's 
> SB> recommended that only 3 or 4 channels being used in one area at a 
> SB> time (i.e. you use channels 1, 4, 7, and 10, and don't touch the 
> SB> ones in between). In the real world, it's more difficult 
> than that, 
> SB> but it's not helpful to have two nodes in close proximity 
> using, for 
> SB> example, channels 1 and 2 for different network segments 
> SB> respectively.
> 
> White paper out recently, due to the bell shape of the spread 
> spectrum usage across the band that makes apparently 5 
> channels useable with the 13 allowable (without overlapping 
> so badly to cause problems)

Ahh, would you consider five being the maximum before things degrade too
badly? I've never seen any documentation relating to what sort of
channel separation is necessary for 802.11*.

> SB> 4. There needs to be one single channel network-wide if 
> roaming is 
> SB> ever to be achieved. Julian Featherston reminded me of 
> this recently 
> SB> (thanks Julian :) and has suggested Channel 1. I concur, as it's 
> SB> both easy to remember and falls under the 4 watt EIRP license.
> 
> untrue, wireless cards can automatically switch between 
> channels without much hassle at all, however network roaming 
> in the mobile phone sense is a little impractical at best... 
> But that's only a present issue due to decent routing 
> protocols etc etc etc

Wireless cards can, yes, with operator intervention. I'm unaware of any
wireless cards which can switch automatically, and if they exist they're
probably going to be running some proprietary software IMHO.

For now, roaming seems to be a proprietary thing (on a vendor by vendor
basis), but 802.11f should sort that out, I hope. Of all the AP's which
support roaming, though, there's a common thread - they all must be on
the same channel and the same SSID. I'm presuming, then, that 802.11f
will follow suit.

As far as routing goes, with forethought it shouldn't be an issue. Have
you seen how 'Mobile IP' works?

> SB> 5. The ACA's "public playground" concept needs to be 
> adhered to, and 
> SB> it's not fair on other 2.4GHz users to stomp all over the entire 
> SB> band available. Nor is it legal to unreasonably interfere with 
> SB> existing users, such as cordless telephones, bluetooth, and so 
> SB> forth.
> 
> Agreed, even some commercial operators are taking this 
> approach that unless we all get along, everyone is set to 
> loose... Some in a very pro-active way (in a good sense, at 
> least on the surface it appears that way)

It would be nice to be able to work with commercial operators on such an
issue, so wireless groups and commercial operators don't stomp over each
other, let alone the public at large.

In regards to one of your earlier e-mails about amplifiers, I agree
entirely that people should limit their output to what is necessary.
Point #5 probably should have emphasised that :)

Thanks :)

 - Simon


To unsubscribe: send mail to majordomo at wireless.org.au
with "unsubscribe melbwireless" in the body of the message