I am still seeing a lot of posts from people getting started in BBHN that are using WRT54G's as their first nodes.  I believe this is a mistake. My suggestion is to start out with a NanoStationM2. Why the NanoStation? Here are my main reasons:

• Ease of getting a node up in the air and running
• Lower overall cost
• Significant improvement in performance over a WRT54G
• Outdoor design of the NanoStation
• More features

I believe you will have a better experience  and be on the air faster if you start with a NanoStationM2 instead of the WRT54G. 

If you want to see more about why I feel this way see the article on my website at:

http://ae5ca.com/?p=74

Clint, AE5CA

Thanks for straightening me out on that fact.  I'm new enough to the Ubiquiti devices that I'm still getting the features of the different models confused.  I guess I was thinking about the Rocket when I was thinking about external antennas. But 13 miles with the built in antennas?  Wow, looks like I better get me a couple of NSM2s!

Steve - N4TTY 

I think there can be some trade off here and cost factors can be rather different than what you have projected. For instance: I currently have 3 WRT54G v2 nodes that only cost me $5 each on craigslist and work perfectly. I can also get a 12db outdoor omnidirectional antenna for under $40 and an outdoor enclosure for as little as $13. So that brings me to a total cost of $58/node using WRT54G. Im not sure I see any real need for POE with nodes but there could be a purpose that I might not be thinking of. If you wanted the nodes to have their own power, as you should, then that is a cost that would still apply to any configuration.

With all of that said, I definitely feel that Ubiquiti has some great hardware over the WRT54G's. In the ideal situation, I would go with Ubiquiti but since I am just starting my ARES team (2 of us right now) all of the money is coming out of my own pocket. But it is clear to see that on a cost perspective there is still a lot of money to be saved using the WRT54G's if you know where to look. If the ability and funds were there I would even go as far as trying to do an 802.11ac mesh. Of course, that being a new technology, we will have to wait for that to get better over the years. I would also have to use a newer version of OpenWRT unless BBHN came up with a supporting firmware. ;)

There is some very good info in your write-up though. Very well done and thought out.

I also have a mix of hardware (LS and UBNT). I think the LS devices have their place as do the UBNT devices. The 600mw output of the NSM2 is great (vs. the LS 100mw at it's cleanest/strongest power).  The MIMO technology in the NSM2 and RocketM2's are really good performance enhancers as well.

In my opinion, the LS devices are good for "light" RF work and tunnel connections (If you can find the rare  WRT54GS models).

The "aimer" script on UBNT is nice for easily aiming antennas. 
The ability to patch firmware in UBNT instead of a full reload is also a big time saver.
I'm currently trying to find a workaround for a nasty bug in the Kamikaze build on Linksys around the package management stuff.  (to support tunnels).   Working with the Attitude Adjustment (UBNT) build is much more elegant/stable.

The lack of multiple ports on UBNT is a bit of a pain, but, not every node needs extra ports.