The location was in a field, less than a half-mile from my residence.   Early in the year several club members did a site survey and attempted to make the connection in one hop of a mesh network. 

This is where it helps to understand more of the context.   The neighbourhood was in the suburb, but in an area is packed with large and tall pine and maple trees.  No tower was available that could get above the foliage.  Needless to say, or initial approach through the thicket did not pan out.  

The internet side of the link was running an outdoor omnidirectional antenna that was vertically polarized.   The opposite side could get a spotty signal in just the right places, but nothing dependable.

We soon scrapped the "one hop" plan.

With field day rapidly approaching we obtained permission from a few neighbors to plug into their outdoor outlets - some as low as 3 feet from the ground.   We placed WRT-54G routers inside of tupperware cereal containers to seal them and plugged them in, following the contours of the road - cutting trees out of the equation.    These containers can be just the right size for the stock antennas, and have the added benefit that they are waterproof, see-through and thus less likely to generate unwarranted concern.   Each had a business card taped to the inside.  A few tests, and we knew that would work.   We left all devices unmodified with original part 15 hardware.

An example is shown below:
Node in tupperware
Once at the FD site, the mesh was daisy chained into the "internet" port of the local wireless router for the site - this provided local access to existing wi-fi devices and hosted the server connection for the field day logging server.

At the site we used the connection for:

* Weather radar
* Software downloads
* Video chat
* Streaming video and music (child entertainment)
* Playing on the mesh 
* Viewing Mesh topology.
* Fixing a member's computer on-site
Through most of the event we were going through two hops before reaching the node with internet access.  We tested the speed to be sustained around 1 megabit in both directions.  Not the fastest ride on earth, but respectable given the number of hops. 

The mesh performed better than expected.   It stayed up the entire duration (except one interruption when it was accidentally unplugged) and bore both the load we gave it and  the severe thunderstorm passing through the area.  

One surprise for us was occasional, but infrequent packet loss on the connection. It was just enough to notice, and we started watching it.  We soon figured out that this was the direct result of our signal crossing a road - every time a car went past, we would see several packets drop - then resume.   The takeaway from this is to mount mesh devices higher over roads.  The mesh protocol does route around problems, but does not respond fast enough to dodge cars at 55 MPH.

All in all, we transferred 5 million packets over the extended weekend, to/from the internet.   Just under 1GB uploaded and 2.3GB downloaded all through the mesh. 

Many club members got to see what broadband-hamnet could do and it proved worthy of further development and consideration.  

My advice to someone starting out: 

* Consider using horizontally polarized directional antennas for long hops
* Long range hops are not common in a residential area, at least without help from elevation changes, tall towers, or bare, line-of-sight paths.   Set appropriate expectations.   Directional antennas can help, but don't expect them to burn through buildings and vegetation even if used on both ends.  
* It is often receive, not transmit, that can be the show stopper.  In my testing I found many areas where I could receive nearly 100% of transmitted packets but could only hear 30-40% coming back.   Location and positioning of antennas around obstructions helps far more - a lot of trail and error here. 
* Watch local community thrift shops for routers - The ones that can run HSMM - configure for that.   The ones that cannot - pair them with HSMM routers but on a separate frequency for device / computer access.   One router used for field day all weekend long was a $5 find in a local thrift shop - the cereal Tupperware container to waterproof it cost more that the router itself...

What it looks like from a topology standpoint - 

If you notice, kc0lql-b is missing on this map, but there is still a partial path from kc0lql-qth to k0smr-2 via k0smr-3…  Routes are managed automatically- the lower the number, the better the connection (“the Link Cost” is lower because there is less packet loss).     


Kc0lql-a can reach k0smr-3 directly - but through walls and obstructions.   It has a high loss, and thus higher link cost.  It would work,  but TCP connections might lose packets and retransmit. 

In this chart circles represent mesh nodes; arrows indicate which nodes can reach each other; and diamonds indicate networks that each node supports.   Internet was provided by KC0LQL-QTH, through K0SMR-3 to K0SMR-2

Mesh node connectivity
All in all it was a great learning experience and something we plan to develop further.
Tim Neu (KC0LQL)
Board Member

Additional information

Close info