I'll try to add what knowledge I know about using WAN
cards for longer distance purposes.
Firstly, other posts clarifies the position with
regard to licenses.
Roughly speaking (please research yourself, starting at say
http://www.air.net.au or http://www.x.net.au) you are allowed to transmit
without a licence below a certain ERP on certain bands, the ISM bands,
around 2.4Ghz and around 915 megaherz. The ERP is determined by the power
of your card, the efficiency of your cable and the directional
characteristics of your antenna compared to an antenna that transmits
'equal power everywhere'.
Basically, by using very directional antenna, and low loss cables,
the signal can go some distance, various reports of some
kilometers, but only at line of sight. (You get the ERP up to as close as
the allowable limit).
There are a number of cards with specialised protocols, e.g. serial line
extenders. The majority are IEEE 802.11, under which there are two types,
frequency hopping (FHSS) and direct sequence (DSSS). The FHSS seem to be
limited to 2Mb/s, the DSSS to 11Mb/s. It seems the FHSS cards are cheaper
in Australia. (See urls above). Be aware, the Canberra site talks of some
older cards available cheaply: I understand these are no longer available.
(also FHSS don't talk to DSSS cards)
The antenna you get with, say, a Wavelan card, is not particularly
directional or efficient, you have to build your own or buy
a special one. The Canberra group had a great time scrounging
old microwave TV antennas and 'converting' them. They also have a design
for a nice helical one you can build yourself with very little
trouble.
Maybe <predator> can bring his to the next slug meeting to show us?
People do use these systems for real live purposes, completely legally.
>From memory, someone was across the road from the ANU. But,
as far as I can tell, the purposes are 'point to point'.
For building a 'net', as in 'anyone to anyone', this (IMO) is a different
kettle.
The latency of a point to point system, as well as routing
practicality leaves a lot to be desired.
Using an omni-directional antenna (you can still get a lot of efficiency
with an omni-directional : you simply don't transmit into the
ground or up into space), you might think a network would
be possible.
Various reports I have read seem to indicate that these kinds on
systems are not reliable.
The problem is due to the 'hidden terminal problem'. Three terminals
A, B, and C. B sees A and C, but A and C don't see each other.
A and C tend to interfere with each other, as far as B is concerned.
802.11 has a CTS/ACK protocol (I think by default turned off) which helps,
(inspired by the MACA protocol, invented by Phil Karn)
but (again, from secondhand) doesn't eliminate the problem.
An example of my pessimistic line of thought is this post
on the Canberra wireless mailing list:
http://lists.samba.org/listproc/wireless/March2000/0119.html
Quoting:
"To build a wide area wireless network, and they certainly can
be built, means starting off with hardware specifically designed for the
purpose.
...
I dont want to put you guys off, but I also dont want to see
a lot of dissapointed people who buy hardware thinking that eventually it
can do something that it really cant."
BTW: the archive of this mailing list is very worthwhile browsing.
To recap: these things are great for line of sight, point to
point.
The technology behind these things is fascinating. Although
the standards exist, you don't have to follow them if it can
be bent for more suitable purposes.
In a similar way, TCP/IP ignored established standards of the time because
they simply weren't suitable.
Jamie
--
SLUG - Sydney Linux Users Group Mailing List - http://www.slug.org.au
To unsubscribe send email to [EMAIL PROTECTED] with
unsubscribe in the text
