Slashdot Mirror
802.11, Horizon Drop-Off And Range
tadghin writes: "Rob Flickenger at O'Reilly Network has written a neat little piece about the range of wireless networks and how high antennae need to be to reach the maximum promised range, given factors such as the earth's curvature and the fresnel zone." Not that most people have solid transmitter disks and clear lines of sight to a wireless reciever miles away, but the more the better when it comes to bypassing modems and expensive per-computer hookups.
I am part owner of an ISP that does wireless service. 25 miles is not feasible from what we've seen. It is possible, but with the stuff we're using (2.4Ghz unlicensed), it's not going to happen. Even with most of the licensed stuff, you're only going to see around 15 miles. The most we've seen reliably from our 2.4 stuff is 10 miles, and this was with an 80 foot tower at the customer location. We also use 5.7/5.3 Ghz equipment for 8mbit point to point service. This is in the UNII band. We have gotten a max of 11.1 miles from this stuff. Could probably get 15 miles with 2-foot dishes.
Hills and things do improve the situation with the equipment that we use with our 2.4 network. The perfect situation is one where you have some sort of blockage just a little before the end of your range. This helps to prevent your transmitters from interfering with their counterparts in case they're transmitting just a bit farther than they should be able to. The best idea would be a cellular style delivery. No reason to go 25 miles with a link. Why not saturate a town or city with transmitters? Build the system (as we are beginning to build ours) where the customer only has to have a small, inexpensive 11db (or lower) flat-panel antenna on their house, without an expensive amplifier or huge LMR-1200 cable.
Wireless is in its infancy, and it's probably going to give us all cancer, but I like being able to take the telco out of the loop (no pun intended).
Instant Karma's gonna get you...
If you check out Cisco's site, they have a nice range-calculation utility that takes this into account.
We've been using long-range 802.11d for about a year. We have demonstrated good connections (5.5Mb) between a 5dBi omni-directional and a 24dBi Unidirectional across 17 miles. However our main tower is 165ft in the air, and has a feed line of 10ft to limit cable losses.
We see a number of installations where the users have put up an antenna, run 200ft of feed line, and wonder why they can't get a connection. A good rule of thumb is 7dB/100ft. For each 3dB your signal losses double, so a person with a 200ft feed line has a signal level 1/16th of the antenna level. You would need 14dB of antenna gain just to recover from your feed line losses.
Basically, if you're trying to run wireless, don't expect miracles. If you play by the posted rules things will work, but if not, don't blame the equipment...
But hills and mountains migth as well *improve* the situation as getting in the way, for the simple reason that all smart access-providers will put their transmitters on one of those high spots.
With a 25 mile range, all you need to do if there's moutains around is put the transmitter at the top of a high one, and everyone who can see that peak, and is closer than 25 miles will get access. Doesn't sound half bad to me.
I'm a lot more skeptical to if the 25 mile range actually is realistical, even taking into account weather and such and not just some laboratory-theoretical limit.
The 802.11 was designed for operation over short distances only. All the timing calculations assume that the air propagation delay is negligable.
If you're going to use 802.11 for outside links, you have to take the propagation delay into account. For every mile between the two stations, there is a >10 usecs round trip delay. The 802.11 standard uses a 20 usec time as a slot length. These slots are the basis for the random backoff procedure, and can also be found in the difference between the various inter frame spaces. When the total round trip delay (air+rx+tx delays) becomes greater than 20 usecs, you'll get (some) performance degradation.
The degradation in DCF mode is graceful, but the PCF will basically break down completely in the face of long delays. Fortunately, most vendors don't even support PCF.
Things get worse if you have more than a simple point-to-point link.
In short, 802.11 can work over long links, but don't bet the farm on it, and results may vary with equipment.
If anyone (i.e. not an electical engineer) wants to learn about radio frequency (RF) propogation, I suggest finding some amateur radio publications such as The ARRL Handbook (2001 edition). It provides plain english explaination and particial experience about the operating in the microwave bands.
Typical 802.11(b) usage is under license-free operation for local "ad-hoc" networks. The equipment is designed to operate locally such as a college campus or a company building, not across town. This relates to the license-free usage exemptions (Part 15 or 11, I believe).
If you want higher power or higher gain antennas, you will need a license from the FCC (in USA) or similiar government agency in your country. Interference with other users of the radio spectrum can result in a fine from the FCC. Story about a wireless ISP being investigated by the FCC.
Although most corporate Wavelan users are considerate enough to not use encryption or passwords to protect their networks, a few paranoid companies have begun to implement these revolting practices.
Don't they understand that networks want to be free?
[Insert the usual disclaimer here]