I oversee communications for the Pony Express Trail 50/100, an ultra-marathon held along part of the original Pony Express trail in Utah's west desert each fall. This area is very remote. Most of it has no cell phone service. There is a 2 meter voice repeater that has good coverage of most of the course, so voice communications are not too difficult, and voice is sufficient for coordinating race operations. But relaying the runners' check in times from each checkpoint via voice to our communications center where they can be uploaded to the Internet is tedious and error prone. We have considered using packet radio, but there are no packet repeaters in the area and most of the volunteers that help with the event don't have packet equipment. So I decided to experiment with Broadband-Hamnet this year to see if it could be helpful in this setting.
I put a mesh node with a 19 dBi grid dish antenna and a 500mW amplifier at our communications center, which is at mile 16 on the race course and is also at the finish line for the race. (The 100 mile runners go out 58 miles and then turn around and come back 42 miles.) Thanks to a 3g amplifier/repeater with a yagi and favorable location, we have fairly reliable Internet access at this spot.
I placed another mesh node 17 miles away at the next checkpoint on the course. This one had a 24 dBi grid dish antenna and a 1000mW amplifier. Aiming the antennas was a challenge. The distance made it impossible to see one site from the other even though there was a clear line of sight. I should have brought a compass so we could look at a map and figure out the the bearings between the two sites. A pair of binoculars at each end might also have been helpful. We started with our best guesses and then adjusted each dish while keeping an eye on the S/N ratio shown on the Broadband-Hamnet web interface. This was a bit time consuming, but it worked and we got a reliable connection established. When night came we turned on a big floodlight at the communications center which was visible at the site 17 miles away. That let us confirm that the antenna there was aimed perfectly at the command center. So adjusting while watching the S/N ratio had worked pretty well.
The laptop at the command center was sometimes connected to the mesh network and sometimes connected to the cellular network for uploads to the Internet. I wanted to be able to upload data from the remote site at any time without worrying about the current state of the laptop, so I attached a Raspberry Pi to the mesh node at the communications center and used it as a file server that the remote node could connect to at any time. SSH would have been the easiest way to transfer files to and from it, but I couldn't find a way to disable encryption in SSH, so I set up Samba on the Pi and we used the unencrypted SMB protocol to transfer files.
The mesh network was used for 24 hours and worked almost flawlessly. There were a couple of brief periods when it slowed down, but I was too cold and sleepy at the time to try to diagnose why. I don't think the antennas had drifted out of alignment because the performance returned to normal a few minutes later. There are no homes or businesses out there, so we shouldn't have had interference from microwaves, cordless phones, or that kind of thing. But there is an army base a few miles away. Maybe they were doing something that occasionally gave us a little interference.
I was very pleased with the way Broadband-Hamnet worked for us. I plan to expand the network to cover more of the course at next year's race.