Need to build a wireless bridge with a 5 mile range. Ideas?

[ShortBus]

I'm quite familiar with using 802.11 APs as wireless bridges, but I've never had a project that requires a 5-mile, outdoor spanse. What type of gear should I look at getting together? Do I just need two Linksys WAP54G APs and a couple of directional antennas? Which directional antenna should I use?

Thanks!

[AdamV]

You need a whopping gain - something like at least 15dBi or more at both ends. The power you need to be getting at the receiving end (after free space losses) will depend partly on the noise floor at each end and on the quality of the radio to distinguish signal from noise.

It might be tempting to use an amplifier to get the power up but the FCC rules make it preferable to increase the antenna gain instead (ie buy a better antenna in the first place). The max power at the intentional radiator is 1W (30 dBm). If your AP produces this power at the radio, you will lose some in the cable and connectors so it will end up below this level.
Now the rules state that at 30dBm you can have a maximum gain of 6dBi antenna (at each end). This gives a nominal 4 watts EIRP, the same as the max for a point-to-multipoint setup. But, in a point-to-point scenario, for every 3dB additional antenna gain you have to reduce the Intentional Radiator power by 1 dBm. (this is for the 2.4 Ghz band, which is probably better over this kind of distance than usgin 5Ghz (802.11a).

so a couple of examples helps:

[darkmagic]

WOW, all that just went over my head

[Jops]

Directional LOS Microwave towers....

[ShortBus]

Eliza, thanks for the great info.

Ok, did some more poking since my OP:

Now I'm looking at using a pair of 3Com OfficeConnect APs (Manu. part # 3CRGPOE10075). I chose this AP for three reasons 1) relatively inexpensive (this is for a church out in the middle of BFE. Cisco gear isn't much of an option, as much as I wish it were) 2) supports PoE 3) has a 18 dBm radio, versus the 15 dBm radios in a Linksys or Buffalo AP.

At one end, I would attach a 14 dBi yagi. The other end would get a 24 dBi mesh reflector grid. I went the yagi route since I figure that trying to aim two parabolic antennas at each other over 5 miles would be a royal pain. Both APs would be placed in a weathertight enclosure and placed as close to the antenna as possible to reduce signal attentuation caused by the coax.

Two questions:
- How workable does this plan sound?
- How can I be certain that a direct LOS exists between the two points (no trees, etc?) Can I rent some sort of laser sighter? One of the locations is a ham radio tower on a hill and the other end is a church steeple, but it's impossible to know visually whether it's interrupted or not.

[AdamV]

sounds reasonable at first sight. 3Com probably a good choice, well done on the research of power output. PoE useful too, especially given your situation (802.3af states +48V DC and up to 12.95W per device. Cisco used to use -48V instead, not sure if they still do, but make sure your power injector is the right polarity!)

your decision about a yagi for one end because it is less directional is questionable (having just checked some references I maybe should not have even introduced them - supposedly they are usually good up to about 2 miles). Don't forget this is a two way link, so it's not just like firing a gun a target, it's also like throwing the target back down the barrel of the gun!

Realistically, you would need to use a signal strength meter to help get the best possible alignment. You are going to be losing something in the order of 118dB over five miles - so you better start with sufficient signal to take that loss and still end up with something detectable (above the noise floor) at the other end. Legally you could use an 18dBm radio with a 42dBi antenna and still be inside the FCC regs (due to cable losses). This gives you 60dBm at the EIRP so that's -60 (ish) to be detected at the other end - a very weak signal but probably do-able.

You also need a compass, a map and a good telescope to check out what obstructions are around the path for the traffic. A hill and a steeple sound like a good high starting point though!

You need a clear "zone" (the Fresnel zone) around the direct path - imagine a very long thin football (the US kind, a rugby ball to anyone else) rather than a cylinder. The radius in feet at the widest point is
r= 43.3 x (d/4f)^.5

(so 43.3 times the square root √ of the distance (in miles) divided by four times the frequency (in GHz - use 2.4, don't worry about the higher channels) Yes, I had to look it up.) For a 5 mile link that would be about 31.25 feet (damn yanks - I must find/work out a metric equivalent formula)

So you stand at point A and check along the path of the signal and decide if anything along that path will be within that radius (0 at each end, swelling to r in the middle). Anything up to 25% of the zone (the whole circle) can be blocked without any major issues to signal loss), even up to 40% depending on other factors.

What lies between the two end points (generally speaking)? Buildings might get built, trees may grow etc. Also watch out for large reflective surfaces (existing flat buildings, car parks, lakes, which will cause problems with multipath.

Sorry if this is going over some heads, the OP seems to know basically what he and I are talking about, so I thought I would get straight to the nitty gritty.

Now, where's mpkn3rd when we really need him to check this with another pair of eyes? I'll see if I can rustle him up by mail...
_________________
"Due to global warming, eskimos now have 20 words for water" John O'Farrell
Meteor IT - Technical Consulting Services and Software Training Courses, Leeds, UK

[ShortBus]

Some raw thoughts:

According to my calculations (and I'm hardly a trig genius, so I could likey be wrong), for each degree of adjustment, the beam will move about 293 feet (89 m). The size of the antenna grid is about 40 inches (100 cm). The reflector grid antenna that I'm looking at has a 8 degree beam width. How difficult is this going to be to aim if I use two of them instead of the yagi?

Haven gotten the GPS coordinates for each location, I now discover that the distance is actually closer to 5.75 miles apart. The church is on a slight hill elevated about 150' above the other location. I can mount one antenna about 30-35' up on the church steeple. At the other location, I can mount an antenna up to 45' on a ham radio tower. It's a rural community with farm land and the edge of a state forest being the main seperators between the two locations.

If I'm understanding what you're saying about the Fresnel zone, if I mount the antenna only 35' up on the steeple, I'd better have nothing but flat land between the two antennas. If the spanse isn't totally bare, I can only have objects no taller than about 12' on the ground between the antennas. Am I right?

[AdamV]

Fresnel zone blockage - sort of, but in terms of amount of blockage you have to think in terms of area of a circle at the point where the obstruction occurs - the top half would be completely unobstructed in your scenario.
As a rule of thumb, a horizontal object (building, forest) which is half the height of the Fresnel radius at the midpoint will be blocking about 30% of the total area of the circle at that point. (draw youself a circle with a horizontal line across at half way up the radius - the 120 degree segment this creates is a shade under 30% of the total area).

At 5.75 miles your zone (for 2.4 Ghz) is about 33.5 feet in the middle.
So if your forest is slap bang in the middle it could have trees up to (35-33.5/2) = 18.25' tall and still only block about 30% which should be fine. If it is nearer to one end or the other this gets better. I guess a "state forest" might be a tad higher than 18 feet though...

This might be an argument for using 5 Ghz (802.11a) because the zone will be much smaller (about 23 feet). BUT the free space path loss (attenuation if you like) will be much higher so you are unlikely to be able to get a usable signal at this distance.

(by the way, I am not enough of a physicist to understand why the Fresnel zone matters, nor why there is no account taken for the beamwidth - you would think an 8 degree beam would have a smaller zone than a 30 degree one)

Yes, aiming these is going to be a real pain! A near 6 mile link is a pretty impressive feat! (by the way at around 7.5 miles the earth itself gets in the way too). Often people will use a slightly lower gain antenna at each end in turn and a signal generator to help with this.

Any chance of a repeater station? I remember reading a "how we did it" website about providing wireless broadband access to a remote hillside rural community - they got the signal there by a series of repeaters with solar powered, battery backed-up repeaters. I'll see if I can find a link.

Other options might include IR laser instead of radio - I have seen this used very successfully for short (< 1/2 mile links) but know almost nothing about the practicalities over major distances.

Sounds like the church might need a collection - gone are the days of "save the clock tower", not it's "get us some bandwidth, dude!"
_________________
"Due to global warming, eskimos now have 20 words for water" John O'Farrell
Meteor IT - Technical Consulting Services and Software Training Courses, Leeds, UK

[ShortBus]

Eliza,

Thanks again for the time you've put into helping me.

I am heading up to the site tomorrow to get a visual survey.

The budget is $1700-2100. This is including both parts and labor, but I am willing to forgo most of the labor charges since I'll turn this into an interesting learning experience and case study.

Out of this budget, I'm looking at spending $300 to rent a handheld spectrum analyzer. I know that it's not absolutely neccesary, but if I'm going to be writing about the project, I want some real numbers for my report. So.... can it be done with this limited amount of funds? The client says he'd be happy with a 1.5mbps link.

FYI, I also stumbed onto this startup company. They make a router with a 30 dBm radio and are supposed to be releasing a similarly powered AP and outdoor wireless bridge kit.

[AdamV]

[ShortBus]

Eliza,

Any thoughts on using a pair of these?

http://www.wlanantennas.com/cpe_2473.htm

Seems like a low-cost solution, but I'm a bit skeptical of it working as advertised (and the "all sales are final" return policy doesn't sit well with me).

[AdamV]

The principle of getting the AP and radio as close as possible to reduce losses is a good one, but I'm not sure about this particular device.

12 miles is simply not possible unless you have very high building or hills with nothing in between - the horizon is at 12 miles around 72 feet off the ground, at this point the earth's curvature gives serious intrusion on the Fresnel zone. This is a claim based on physics and maths, not real-world geography!

The numbers do stack up on paper for your 6 mile link - 41dBm transmitted would be possible to pick up at the far end as they have good sensitivity at even the higher speeds.
Don't be fooled by the 18 degree beamwidth - this does not give you 9 degrees of tolerance as the central few degrees will have a concentrated signal, so you still want to align them as accurately as possible, ideally using a scanner. (NB: don't stand too close to the antenna when taking measurements, certainly don't hold it - the signal pattern will be very distorted by the water in your body. Stand behind it (or below, if on a ladder) and ideally a couple of yards away)

Also, the waterproof radome enclosure is great for weather proofing, but the signal direction will be much more subject to wind problems than a mesh antenna (by design).
_________________
"Due to global warming, eskimos now have 20 words for water" John O'Farrell
Meteor IT - Technical Consulting Services and Software Training Courses, Leeds, UK