WiFi over distance - 300 metres

I assume you have lots of money to spend. Nobody gets 108Mbits/sec thruput from a commodity cheap wireless bridge. That's the connection speed. The thruput is about half or less of the connection speed. You'll also find that 108Mbit/sec works just fine out to about 2-3 meters. Any more requires big antennas. At about 20 meters, timing becomes a problem and the bridge will slow down. Basically, Turbo-G and Super-G are not suitable for 300 meter links at full speed. They work just fine for slower speeds, but not at speed.

Methinks you will need to specify a slower thruput speed that is acceptable. At 300 meters, a pair of cheap bridge radios (Linksys WAP54G) with 24dBi dish antennas will work at 54Mbits/sec or about

25Mbits/sec thruput. The problem is reliability. NOTHING on 2.4GHz is considered reliable because of interference issues. All it takes is one leaky microwave oven, a wireless security camera, a municipal wireless deployment, or a nearby 802.11g network, and your reliability is lost. If you truly want reliability, I suggest a licensed wireless solution, with frequency and location coordination to insure that there is no interference.

I suggest you revise your specifications to include:

1. Desired thruput (mbits/sec)?
2. Full or half duplex?
3. Total number of MAC addresses that must be passed?
4. Do you have AC power on the rooftops?
5. Distance (300 meters)?
6. Fresnel zone clearance at midpoint?
7. Maximum cost?
8. What country?
9. Any local interference on 2.4 or 5.7Ghz? Consider 24GHz?
10. Exact reliability figure.

The last item is critical as it determines the system fade margin. Fade Margin Reliability 8dB 0.9 18dB 0.99 28dB 0.999 38dB 0.9999 48dB 0.99999

99% reliability may sound good, but that means your link will be dead for 1% of the time or you will experience 3.7 days of outage every year. Tradition and experience says that it will happen at the worst possible time. Increasing the fade margin by 10dB (one step on the above table) is not a trivial or cheap exercise as it requires either a 10 times increase in transmit power, or a 4 times increase in antenna size at both ends of the link. Pick your numbers carefully.

I you just want to look at possible products, see: |

| is a distributor that carries a wide range of wireless bridges. If you call them, be sure to complain about them hiding their prices. Also: | don't want to make any recommendations until your supply more detailed requirements and limitations.

Good idea, but very expensive. The LED variety is adequate for rooftop work at 300 meters.

problem is that at 100baseT speeds, the prices for WB4100 boxes are about $8,000 for each end. That's about as cheap as it gets for

100Mbits/sec.

I've only done two FSO systems. One of them works just fine with no problems except when the fog rolls in. The other was the nightmare from hell. It was across a freeway at 850ft and an elevation of about

50ft off the freeway surface at one end and 20ft at the other. It worked just fine most of the time, but would constantly fail in the mornings. The problem turned out to be that the inversion layer created by the vehicles heating the air at the freeway surface caused the beam to be diffracted. I measured a beam shift of about 6" at one end. Then the inversion layer was finally broken up by the morning sun, it was replaced by sufficient turbulence to again lose the beam. I spent about two weeks trying to salvage the sale and eventually gave up. It was replaced by a 5.6GHz link which worked as expected.

I've also done some troubleshooting work on FSO. One system would fail consistently when it was hot, but work otherwise. I found that one end had been mounted so that it shot over the office building HVAC unit. Every time the air conditioner fired up in the summer, the heated air turbulence trashed the beam. Moving the FSO box solved that problem.

Another more serious problems is the stability of applications and services that fail to appreciate having their network connection disconnected. I've seen far too many crashes and hangs caused by ancient database applications that have no tolerance for loss of connectivity. Drop the LAN connection on one nightmare retail (cash register) application I have to deal with, and it locks up. That's because it opens a new socket number on reconnection leaving the old session hanging.

Moral: Test your application before you bridge.

How about looking at some of the free space optical stuff?

This is probably overkill for your 300m mind you but you said you wanted fast, it's faster than 108Mbps :)

David.

The original poster offered no budgetary constraints, only that it should go as fast as possible. :)

David.

Well, in that case, bring on the big ticket laser FSO boxes: |

GigaBits/sec to 1000 meters. Only $27,000 per end. Don't forget the RF wireless backup in case of fog. 60Ghz, 480Mbits/sec. | $10,000 per end. If it rains, the 60Ghz will go down, but the FSO will run. If it gets foggy, the FSO will go down, but the 60GHz will continue to run.

Incidentally, there's always a cost limit. It's most obvious immediately after the initial sticker shock. I usually ask my customers "How much do you wanna spend?" and then tell them what they can get for the price. Saves quite a bit of negotiation wear and tear, and tends to give a better picture of reality.

I'm a big fan of do it thyself:

that doesn't offer much reliability.