FabLab helps the developing world set up long-distance wireless Ethernet

The wooden frame seen above hosts a parabolic reflector making up one side of a wireless network link. This is a Fab Lab project called FabFi which uses common networking hardware to setup long-distance wireless Ethernet connections. It’s a bit hard to tell in the image above, but the reflector focuses radio waves on the antennae of a router we’re quite familiar with, the Linksys WRT54G. It’s held upside-down in an enclosure meant to protect it from the elements. The node above manages to complete a connection spanning 2.41 miles!

One of the core values of the project is to develop hardware that is easy to build with limited resources, then to make that knowledge freely available. Anyone who has the ability to download and print out the 2D design file can build a reflector for themselves. As we’ve seen in other projects, paper stencils and hand tools can handle this job with no need for a laser-cutter (which was used for the prototype). WRT54G routers are inexpensive and the project uses the open source firmware OpenWRT. They can be run from 12VDC power which means a car battery works when mains power is not an option. The system has been running in Afghanistan for two years and hardware failure is still in the low single-digits.

[Thanks das_coach]

29 thoughts on “FabLab helps the developing world set up long-distance wireless Ethernet

  1. Simple questions: I read somewhere that basic WiFi over long distances degrades quite a lot.

    This is the longest I ever heard of, and using off the shelf components too. Quite impressive.

    What’s the max range?
    Can those old large satellite dishes be used? What of the small dishes?

    Also, can this be used with Mesh concepts? I would imagine that a system running pfSense with 2 wireless routers, one interface a WAN (uplink) and the other treated like a DMZ (down), each to IP Filter the other out so it does not “talk to itself” though the two wireless routers and a single system could comprise mesh network with off the shelf parts, and combined with this antenna technology, may be a good answer to these “internet kill switches” and centrally controlled social networks.

  2. damn I’m trying hard to troll this, but this is tough. Good read for sure. It makes me happy to see good open ended projects come to life and not get polluted with political and monitary factors along the way. This seems really useful for people who live in buildingss where verizon corporation/ATT/cablevision/comcast refuse to run highspeed lines. My father’s retail property is one of these places. Verizon refuses to wire internet cables from across the street since it would cost them upwards of $100,000 to relocate a stream or something… Perfect solution here.

  3. We’ve been able to do stuff like this and meshing for quite some time. Once in a while I hear about long distance WiFi being used, but I don’t ever hear about mesh being used. It seems like a great idea, but I suspect people aren’t using it, because those who have internet connections don’t want to degrade them with additional users and those that don’t have internet connections who try to mesh find that there are no other mesh nodes to connect to.

  4. @Doktor Jeep — While sat dishes would definitely improve the range and throughput (assuming you properly tuned them), I would suppose that in Afghanistan that chicken wire is much more plentiful.

  5. @Doktor Jeep

    Signal can definitely degrade over distance, but it’s fairly predictable if there is direct LoS (Line of Sight).

    This is actually not even 1/100th of the record achieved with off the shelf hardware…which I believe is 382km (238 miles). I’ve consistently gotten 20+ mile links with a simple Parabolic dish and/or Yagi antenna.

    Ubiquiti Bullets can be used to get 20+km links with relative ease.

    It isn’t my expertise, so take the following with a grain of salt. Meshing is definitely less about distance and more about discovery and routing. I think distance would be much more reliant on output power + antenna design + LoS.

  6. @wifigod – -agreed… There is plenty of optimization that can be achieved in this design, even with local materials. My first change would be to hang a cantenna at the chicken wire’s focus point. The 2nd change I would make would be to elevate the entire assembly up about 1000 feet to increase the line of sight.

  7. Great story, I’ve tinkered with this stuff just across my town with a friend using a router a directTV dish and a pringles can. (we were able to connect… that’s about it.)In these days of ever tightening security and impending privacy rape-age I’m surprised more people aren’t investing time and thought into a home grown wireless network that can’t be brought down just because some suits feel it should be. Maybe the real wireless revolution will start in the third world?

  8. 1) As stated above, much longer ranges are easily doable. I had a reliable 20 mile 30Mbps link set up using two 3′ parabolic dishes for a while…

    2) Most WRT54G’s will run on anything between 5 and 35 Volts if you swap out the input filter capacitor…

  9. Amateur (Ham) Radio people are already doing this, AND with MESH, AND with WRT54G’s. Hopefully soon ported to other hardware, now that one gent has figured out the ‘trick’.

    I personally have shot a 6 mile link (100% Link Quality) across South Austin TX between 2 parking garages, with 24dBi BBQ dishes. Lotsa geeky 2.4Ghz noise in Austin with all the personal access points, microwave ovens, 2.4Ghz cordless phones, etc.

    The problem with bigger dishes is along with the higher gain, you have a smaller beamwidth. Think of looking through a telescope–the more powerful the scope, the more precision you have to aim and keep aimed.
    Just a little bit off and you lose the link. We had some problems that day because of the 25MPH+ wind knocking the BBQ dishes around. They are only about 2′ x 3′ with big gaps. A 6/8/10′ AL mesh TVRO dish would have to be very securely anchored to keep a steady link.

  10. To all those saying “LOLOLOL I CAN DO THIS WITH $300 AND SPECIALIZED TOOLS”, please note that this is being done in Afghanistan where you can’t exactly Ebay your way out of the problem.

    The notable point here isn’t the distance achieved but the ease of construction. That said, of course there are improvements that can be made.

  11. Highly directional antennas most likely are incompatible with a wireless mesh network. For the network to be about to route problems all stations will have to hear at least one other station. directional antennas would be use for long distance interconnects between networks, or allowing an end user connect to a node in a network. From what I’m reading that’s what the features antennas are being used for. While the goal is to provide internet access, there is no mention of free internet access. Looks as fabfi is being tested in the USA.

  12. What did all of this cost? You can buy a pair of Ubiquity NanoBridge M5’s for $160 USD and they will link two spots over miles easily. Comes with all the hardware you need, minus a set of antenna towers to mount on for the really long-distance stuff. And its a commercial product with real support.

  13. I’d build a simpler tower and make a simple Yagi out of whatever conductor and insulator was widely available locally. Use whatever radio you have or can get easily, as long as it has a detachable antenna. You can still use (small) paper templates for most of the cutting etc. I don’t think this would require anything more difficult to acquire or build than their design, but is smaller, more flexible, uses less material and should require less build effort. Seems to me that cutting wire to the right length is easier than cutting odd curved shapes out of wood.

    Seems to me that ‘single digit’ failure rate is pretty awful, though their sample size is probably too small to mean anything.

    Anyway, forget all that. These guys are out there in Afghanistan actually getting shit done and making things work. Mad props.

  14. I have set up a 27km wireless link using wrt54 hardware with an off the shelf 25dbi grid antenna running freifunk. the schlep of trying to make an antenna while ubquiti makes a 22dbi dish with integrated router selling at 87USD in South Africa negates the small saving when compared to the wrt54.
    The ubiquiti also includes MIMO and at 7km I get 140mbps full duplex.

  15. Yeh, I was one of the guys who did 125 mile link with 802.11b, 2.5 miles is pretty easy to do with both homemade and fairly cheap commercial gear, but that doesn’t cheapen what they are doing. They are trying to create a whole network. With interfering stations and critical links on normal topography. When we set our record we went from mountain top to mountain top, which makes things a lot easier. We also didn’t limit ourselves to a $65 budget. We spent about $1500 on materials alone. I do really like their parabolic reflector. Wind loading can be a significant problem with a lot of dishes, their system seems to eliminate that. Looks like they could get better performance out of it though. The feed point they are using is just a stock folded dipole. Most of the energy radiated from it will not hit the parabola in the way they graphics show. The should really look into a better feed, such as a patch antenna. I would guess they could get a 4x to 10x performance gain with a better feed with their same reflector. Great job guys and good luck!

  16. I haven’t heard any mention of Fresnel zones… Important when you’re talking wi-fi on great distances. But ya, the tech is old news, but using it to give people access to the internet they wouldn’t otherwise have – great jerb!!

  17. Fresnel zone, meh. My experience with long-ish range wireless is simple:

    If you can see your target, and there is not too much noise, it will work.

    I’ve had a 17-mile 5.7GHz link running for about five years now over mostly flat terrain with trees. Even with plenty of height (one end up around 150 feet, the other around 60), the Fresnel zone is rather completely trashed, and it doesn’t seem to really matter: You can literally see between the two points.

    On the other hand, I’ve failed at 5.7GHz over distances of just a few miles because of a few trees obscuring line-of-sight. (2.4GHz worked acceptably, despite obviously having much higher background noise.)

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.