Slashdot Mirror
Can 802.11 Become A Viable Last-Mile Alternative?
NikiScevak writes "As telco's around the world move from government hands to private investors the incentive for them to create compeition at the wholesale DSL level drops dramatically. The CSIRO in Australia are investigating the use of Wireless LAN technology 802.11b as a means through which to provide alternative broadband access, achieving range of up to 7km with standard components."
There are several isp's selling wireless access for the last mile in North Carolina. Overall, I wouldn't touch it. The networks are generally insecure, sniffable by anybody and their palmtop with the right hardware/software. From what I have seen and heard from people is that it works, but some days it dosen't work as well as others. *shrugs*
Honestly, I wouldn't mind being able to drive around and have allways on access in my car or something like that, but wireless does not cut it.... Collissions, and cordless phones reek havoc with 802.11b. I use a 100mw ap at my office... when I'm on my cordless phone... my laptop says the link quality is 10-20%.... and the ap is 20 feet away...
But try this in the valleys of South Wales and you'll soon realise that copper has its advantages.
That's very incorrect. Directional antennas won't
help that much from interception and interference. You will still get the signal
out of their projected beacon (which is still several degrees wide, BTW),
but a bit lower. Radio waves don't work the same way
light does, it's like thinking that nobody will hear
you shouting when you go behind a building..
consume.net in the UK are pioneering this kinda thing. There's also a whole raft of other community based wireless links at Wireless Anarchy.
Al.
HantsWireless - Hampshire Wireless
SurreyWireless - Surrey Wireless
The limit on a Cat 5 run is around 100 meters. A mile is 16 times that. Also, Cat 5 is meant for indoor use. You need to count the appropriate outdoor conduit in that cost, as well, which in many cases may be several times the cost of the cable itself (just like how it costs hundreds of times more to dig up the ground than the fiber to put in the hole does..which is why we have so much dark fiber. As long as they've got it dug up, they put in as much as they can afford).
Then you've never been to Europe lately? Here, we have a decent GSM-network that almost never fails
GSM has an intrinsic part of its design to ramp down the power that the phones transmit at when the signals are strong. It was always designed to work in a crowded network. After all, it has a 35 Km range in its design, yet a cell in the centre of a city would theoretically cover most of even a large town.
This was one of the biggest problems with older analogue networks - they always transmitted at full power and had trouble with crowding out in densely populated areas.
As a bonus, your phone's batteries last alot longer in a city than in the country on a GSM network (but not on analog phone).
Yours,
Michael
There is no cryptographic solution to the problem where the intended receiver and the attacker are the same entity.
At the provider side? For a few point-to-point customers yes, but normally you have an omnirange at the provider and (more or less ugly) directionals at the customer side.
I live in Slovakia, where there still is a monopoly for the wired local loop to the end of this year. We have no commercially available DSL yet. Of course the wireless is cheaper and everyone and his brother is using it for everything and does not give a sh*t about the regulations.
The band already is clogged in the bigger cities. It does not matter how one company plan the network - there are many and they are not going to plan it together.
The reach is no problem - I know of a few 20 km point-to-point links. The density is and the unregulated band is not a way. There are technologies in the regulated bands (FWA at 3.5 and 26 GHz here) that are meant to provide a high-speed local loop. WiFi as a last mile is a kludge - it will work but...
With proper antennas, you can hit 8 miles or even more. Read more about it at DSLReports in the wISP forum at:
http://www.dslreports.com/forum/dslalt
Somebody once told me that a T-Bird (T1...T3 packet sniffer) cost 40 grand.
Yeah, and a PC used to cost $4G, too.
Dallas Semiconductor makes E1/T1 framer ICs which you could interface to a Motorola 68k or something nice and fast for peanuts. It's been a while since I went through the Dallas datasheets but I'm certain that you can use them to sniff the data stream with a little extra circuitry to block any transmissions from the third (sniffer) framer. The actual data stream on the wire is very well documented and if you put something like this on a PCI card and modified some Linux WAN drivers I'm sure you could make a sniffer without too much difficulty. Hell it'd be even easier if you modified an existing supported WAN card with an internal DSU, like the LMC 1200.
No matter how you look at it, it'll be hardware mods + software mods, unless the framer can be programmed NOT to emit anything, which I'm not sure is possible. Also DS2/3 sniffers will be a good sight more expensive I'm sure. The loop lengths on those are not very long for copper and there's a lot more critical timing.
Now you could say that this knowledge is specialized and that the design of such a thing could be $40k -- true enough. I happen to have the knowledge and I do contract design work... :-)
Twisted-pair ethernet uses differential signaling (a transmitted "one" bit is sent out as a positive pulse on the TX+ line and a negative pulse on the TX- line). There is no requirement for a common ground.
It is entierely possibal for your comptuer to be at 100 volts realtive to your neighbors.
No, because the ground on both computers is plugged into, well, the ground.
it will destroy your computers.
But what won't these days?