There are only two affordable options.  One is satellite, the other is fixed wireless.  Satellite has a couple of drawbacks.  It is expensive and has a high latency.

This is a story about fixed wireless and how it has been deployed in rural Nebraska and Colorado. 

Population density of 1 person per square mile.  10,000 square miles, 10,000 people.  50% live in 3 small towns.  The largest has a population of 2,000.  Most live on farmsteads surrounded by elm and cottonwood trees.  The region is flat with a few 100 foot dips.  The highest buildings in each town are 200 foot concrete grain elevators.

In 1999, we applied for a grant from the Rural Utilities Services, asking for help in building a fixed wireless network to serve these communities.  RUS provided $350,000, which was matched by another $350,000 by the ISP and local recipients.

This money was used to buy 10 access points (each connected wirelessly to the main office), 225 computers for those who didn’t have one, 400 client connections, and a dozen connections to places like the community center, library, hospital, etc. 

Prior to building the network, a wireless engineering outfit was hired to perform a site survey.  They sent out two employees to scout out the area.  The “engineers” brought out a spectrum analyzer and a GPS unit.  We tagged along, wanting to learn everything.  They were here for two days, but could have done the job in ½ a day.  They didn’t appreciate our watching -- probably because a site survey is a pretty simple procedure, even though it is essential. 

A week later, they reported our plan would not work.  So, what do we do?  Give the money back to the government?  We talked to the salesman.  He assured us the radios will do what we want.  So, he sends us a couple of radios with which to experiment. 

We had some major questions.

• How far would the signal go?
• Would it work through trees, rain, fog, snow, wind, etc.?
• Would a bunch of 2.4 GHz radios in close proximity scramble each other?
• How many people could access the network before crashing it?
• Would people want it?
• Would the radios be stable?
• Was the budget we started with realistic?
• Would the grain elevators let us use their facilities?
• How may of our dialup accounts would be cannibalized?

 At the time, there was no one to ask.  To our knowledge, there was no one doing a large scale deployment to residential homes or between communities.

To answer these questions, we clamped a test radio to the railing on top of the elevator.  We hopped in the car, stuck an antenna out the window, and fired up our laptop.  We drove around the region while streaming music off a remote machine.   It worked.

Without anyone to answer the questions, the only thing to go on was our drive around test, the no-go by the “engineers,” and the "yes" by the salesman.  Based on this, did we want to gamble our life savings on the idea?

Yes.

Deployment

So, we hired the next-door neighbor kid who had just finished home school high school and set about making it happen.

1. Obtain leases from grain elevators, city water towers, private radio towers and grain legs.  (In all cases, we did this in exchange for access to the service)
2. Modify the budget to reflect reality.  (More money for backbone and less for each client.)
3. Find money to purchase equipment.  (We have to buy before the grant reimburses. Banks refuse to give a bridge loan, so shuffle money around.)
4. Purchase equipment.  (Computers incompatible with wireless cards, find a new vendor)
5. Install backbone.  We made our own masts, aligned our own antennas, and learned to climb and tie knots.  This went smoothly.
6. Install some clients to make sure it works.  (First installs were friends, in order to make sure someone would be understanding while we worked out the bugs.)
7. It works.  Continue to install backbone and clients until we finish.

Today, we have about 400 homes and a few businesses connected. 

Answers to some common questions. 

• We use 2.4 GHz unlicensed radios.
• We use Lucent (Orinoco) access points and silver cards at our high points.
• We use silver cards at the client location.
• All radios are Spread Spectrum, not Frequency Hopping.
• Our longest point-to-point link is 23 miles.  We could go further.
• Max range on the point-to-multipoint links is 12 miles.
• We use 8 of the 11 available frequencies at our main site.
• We do not use encryption.  We do use VPNs on sensitive links.
• We charge a $250 setup fee and charge $25 a month for unlimited access.  We reserve the right to limit bandwidth to 128k, but have given people unrestricted bandwidth.  Average person sees a 350kbs connection which bursts to 900kbs.
• We own the client equipment.  If the client leaves the area or stops paying, then the equipment is returned to us.  There have been three returns.  Two because the client died and one because the client moved away.
• We provide the client with a NAT address .

• We allow clients to network their computers and share the wireless connection.

• We route unauthorized cards to never-never land.  We have found no unauthorized cards.

• We can run about 60 clients off of one access point.

• Wind, rain, snow, fog, cold, heat, and ice do not affect the service. Hills and trees do.  Metal buildings do.

Does it work?  Yes.

Problems

• Most of the problems have come from the backbone radios.  Certain models were not stable and were replaced.  If the radios (the shielded radio chassis) are placed too close to each other, they will interfere with each other.  Of course, this was news to tech support. 

• We have had 2 omni antennas shattered by causes unknown.

• Half the town could get a signal.  The other half could not.  Found the omni antenna parallel to the ground.  Bolts had come loose. 

• Two of the omni antennas filled with water.   Haven’t figured out how that happened.  Stick one in a bucket of water for 10 minutes, and it wouldn’t fill.  Yet, 220 feet in the air, it did.

• One point-to-point link will occasionally fail on Sundays.  We have a redundant radio in place to cover for the failure.  Black helicopters are suspected.

• One client machine struck by lightning, one shot with a shotgun, and about 10 silver cards destroyed through various acts of "wisdom," including application of foot pressure.

• We have 20 backbone radios.  Five failed while under warranty.  Three have failed outside of warranty.  Extended warranties are not available.

• Customer calls and is without signal.  Software checks out.  Says antenna okay.  We ask, "Is there anything between your antenna and our tower?"  They'd parked their semi-truck between.  Called month later without signal.  We immediately asked where truck was parked.  They found different parking spot.

Comments

I think it is because of the nature of the community that we live in that what we have done works.  Only 30% of our customers have used Napster.  We have college age individuals in residence only in the summers.  We have very few businesses, as this is a very agricultural community.  We do not allow servers on the client end.

We use Spread Spectrum instead of Frequency Hopping.  For this market, it is the correct choice.  For other markets where there is competition, Frequency Hopping would be better.  It is much more expensive.

We have learned a lot.  We know more than most about what works and what doesn’t.  What would we do differently?  About the only thing we would change is the backbone radios.  We have had quite a few problems with them.  We love the Lucent/Agere/Orinoco equipment which provides most of our point to multipoint and client equipment.  It is stable, reliable, and inexpensive.

What we have done can be replicated elsewhere.  Costs have come down by about half since we started.  It will work in places that have hills and trees and buildings if you place your antennas correctly.  

Would we do it again?  Yes. 

Questions?    steve – 308 882-3000