Mesh Networks for Rural Connectivity?

pa_jones asks: "This recent article (author approved transcript) describes a potential low-cost wireless solution to rural/last mile internet connectivity in an Australian context using modified forms of Nokia's Rooftop mesh topology wireless routing transceivers, and low cost Cybiko hand-held gaming devices. I would be very interested to hear the community's opinion of this proposal in general, and specifically about the price effect of operating spectrum, data capacity, transmission distance and the complexity of the operating system required to drive the wireless routers/terminal devices/CPE. Given the collective ownership aspect of such mesh networks and their possible utility in rural and developing countries, I am also wondering whether this type of networking device and its operating system might be a possible joint project for the open source software and open core hardware communities."

Funny Looking tree there Bob

[gavinhall]

Posted by deuxdrop:

The rooftop arials (?) are not always the most attractive... and there are some concerned with the asthetics of their domicile. personally, i would like to have the technical aspects, but the mere sight of it is unpleasent. I would not want to see one on the neighbor's roof (of course in rural Austrailia, your neighbor might be 12 miles away in which case it would be fine)

I honestly don't know...

[Zwack]

But I'd love to see some sort of low cost wireless networking out there.

If it was cheap enough for rural/developing countries I'm sure it would be used in cities as well. Who wouldn't be willing to run a small local node that could talk to other local nodes and act as a base station for mobile nodes within the immediate vicinity?

I know that I would run one of these if I could.

Ham's APRS

[jon_adair]

Anyone looking to do this sort of wireless mesh should definately take a look at how ham radio operators implement APRS [aprs.org]. A lot of the same issues would have to be addressed. You don't want every station relaying traffic, just the ones with good, high antennas. In APRS, stations with better coverage are designated as "relays" or "wides".

It's actually pretty amazing when enough stations are in an area for it to work, even though it's usually 1200-bps with huge latency. Hams also do odd-ball things like routing packets to and from the internet.

WaveLAN / Antenna Solution

[PatJensen]

If you are looking to provide a cool metro-area network solution, Lucent WaveLAN cards can be extended to some distances of up to 11 miles. It works great when you are in line of sight of your destination.

There are a lot of people looking to build out MAN's in their cities, with some in Seattle and San Francisco already running. You can tap into them and get free Internet access and share files with your neighbors.

Check out the following links:

• http://www.wavelan.com [wavelan.com]
• http://lists.samba.org/pipermail/wireless/2000-Apr il/001371.html [samba.org]
• http://www.notwise.net/LUGOJ/2001/Jan/0764.html [notwise.net]

-Pat

It doesn't survive math and human nature...

[human bean]

I can tell you all about stuff like this, having worked with it for some time.

Mesh-routed networks are reliable as long as there are no "evil" (misrouting, misrepresenting) nodes within the network. The one thing that brings them to thier knees is a duplicate node address or mangled routing information transfers. This was not a problem for the military, after all, they had control of the manufacturing process (being the only customer) and relied on other means of radio security (spread spectrum, interrupted carrier, etc.) These things will get hacked (the first few I saw installed disappeared off the rooftops after about a week) and there isn't going to be much you can do about it.

Second, what happens when everybody in a neighborhood has one of these things? Spread spectrum is a wonderful thing, but it relies on the time vs. frequency graph being relatively sparse. Putting a bunch of them in a small area is the absolute worst thing you can do to a spread spectrum radio. Case in point: Local electric utility installs 900 MHZ spread radios on every electric meter in a subdivision. Runs a truck through subdivision to key the spread transmitters and read meters. Not only did they have to drive truck at one mph in order to get all the meters (effective throughput rate of like four baud) but folk's cordless phones are hosed, and the local fire department emergency radio LAN goes down at same time. Eventually fixed, but shows the need to properly engineer such things. Who will coordinate and provide this?

Third is safety. Consider what happens when folks start stretching these things out to the limits (array antennas, etc) in order to cover rural or sparse suburban areas. They are bolting a microwave transmitter up to the side of the house and then instead of allowing the microwave energy to radiate normally, they concentrate it. toward the nearest town. Not much will happen to the owners, I grant you. I just don't want to be a member of the group that is blasting stumps in the field next door. Granted, the power is low, and so is the chance, but the potential for damage is high. Why do you think the road crew has the "no cellphones/two-way radios" signs?

I think the concept is a real good one, but I fear the implementations need improving. I have toyed with a high-powered infrared sort of thing like this (effective even through snow at shorter distances)and using a sort of crypto "buddy" system to eliminate misbehaving nodes (all routing information is held crypto from the user, and a node has to authenticate itself to one or more other nodes to get the current keys. If neigboring nodes detect misbehavior, they refuse to route or accept that node's traffic. Not perfect, but I'm working on it.

This like all the others...

[windowsLuser]

Has great possibilitys, but with true monopoly telco's and cable companys the only current viable Broadband options out there. This has no chance unless they (the monopolys) decide to deploy it. For rural Australians, it sounds great though.

opencores.org not .com

[pa_jones]

My mistake - sorry. Perhaps a more direct link to the rationale of open core hardware development is this link [opencores.org] (http://www.opencores.org/OIPC/why.shtml).