Hi all,
I'm thinking of making a mesh that covers about 150 miles, using a dozen 200ft ham towers (one 400ft) around the area.
Area is central Florida so very flat. Existing vhf repeaters at this height have about 20 mile ranges.
Towers are spaced 10-20 miles apart, the hope is that we could make a mesh that would cover a good chunk of central Florida that "all we'd need is a small dish to aimed at a local tower and be on a large network".
In my professional work I played with a 20 acre wifi project using TP-link CPE710's (as the AP's) and looks like they are just getting added to AREDN.
The beamwidth width is about 7 degrees with their reflector ears but without them seems to be about 45 from my experience, I'll need to do a test setup on a tower to verity their angle and distance range.
So here were some ideas.
Initially, put 4 or 8 AP's in cardinal compass directions on a tower, running a ethernet cable from each to ground, to a smart switch and local router.
That way all the expensive hardware is on the ground and can be easily changed as needed.
Alternative would be to put a smartswich up at the access points and run a power and lan cable down to the ground to the router.
Originally I was thinking of doing as an infastructure setup, with each tower being a large bridged AP with dedicated uplink dishes to known towers, but...
then we have AREDN that might be more useful since a tower may be able to see others and offer multipath redundancy.
My concern is that if the backhaul and users are sharing the same 'sectors' those speeds will default to slowest user.
For ID's I was thinking of using their grid square and direction. XXX-EL98fw-NW
If there was a second SSID that was common, users could hop from sector to sector and tower to tower? XXX-Common
Could ARDEN handle the hand-off, I know it wont be like cellular but if the break is short enough downloads may continue.
Thoughts?
I know some users have a disdain for TP-Link, but they are moving in to the pro market with their Enterprise line and Omada cloud.
(practical clone of some ubi equiq) this units rating is 23dBi , -27dBm
Ken, KN4MDJ / wb4sa
Locally, by mid 2017 our area's attempt at building a 'mesh topology AREDN network' was a bit of a failure.
I mean building the network was a success. It was having it function as a network failed.
Failure:
There were 32 nodes on channel '-2'.
Many hidden nodes and many exposed nodes.
As if any of each are not already too many. ;-)
Our sole 'node array covering 4 cardinal directions' has never had a user to link (ever, in over 8 years).
This system was rebuilt recently. Still not a single user has linked.
All downlinks installed with the expectation of
'if you build it, they will come'
has, to date, failed miserably.
Success:
High profile sites are now connected with point-to-point 5 GHz links.
All high profile sites with channel '-2' devices have their ability to link to each other on channel '-2' disabled.
i.e. Only user access on channel '-2' is encouraged.
We installed 5 GHz downlinks from high-profile sites toward existing/interested users.
Often the user provides devices with enough power and gain to achieve the link.
When you say 'cover 150 miles' is this linear?
I recommend using a tool to determine equipment needed to establish a link between the towers.
I recommend that you do not share user channels with backhaul channels.
When you say "For IDs" do you mean Node Name or SSID?
I recommend that users do not 'hop'. Actually, they cannot access services by 'hopping'.
I recommend that you build your network so that the AREDN firmware may/can/will be able to choose a valid path.
I recommend that you do not install nodes in 'willy nilly' fashion in the hopes that the
AREDN firmware will 'figure it out' !
I do not understand 'hand off'.
I recommend that you build your network in 'cellular' fashion.
" -27dBm"
Typo: dBm values are always positive. negative power ?
Your millage may vary.
I hope this helps,
73, Chuck
Yes, linear 150 miles from furthest points, a sweeping 90 curve across the state from Tampa to DeLand.
So your advice would be to put AREDN on each tower for the clients to connect to, then use point-to-point, maybe bgp to connect the backhaul between towers (setup for multi-path uplinks). Sort of like our current SarNet (microwave backhaul with linked uhf repeaters on VOIP).
Was referrring to the SSID for the naming scheme. Still have to do equipment testing to see what the actual hardward range is for the point-to-point, and see if it's posible to link each tower.
I would say for the tower to tower, use something like a dish (LHG-series from Mikrotik or maybe one of the Beams from Ubiquiti). I don't have experience with any TP-Link dishes, so I can't give you a recommendation there. The alternative (albeit expensive) is to put three Rocket M5's (or most likely the M5ac series now) on each tower, with 120-degree antennas. That would give you the envelope around the tower, as well as potentially the links between. You can go to http://link.ui.com and set up different products from them to see what the bandwidth and capabilities are. If you want to use TP-Link, pick some of their Nanostations on the site, as those are going to be close to what the CPE-series looks like and has for capabilities.
I'll echo the avoid 2.4GHz nodes. And I would say get the ac class of nodes. Above all, plan it out before you buy or put up anything. We learned the hard way that not researching before we send a climber up is an expensive mistake to make.
Have a great day. :)
Patrick.
Unsure about TP-Link gear.
Orv W6BI
It might be helpful to reach out to Gerard Hickey (WTØF) and/or Sean Hage (KD4WLE). Both are leading the way building out the MESH in East/North Cental Florida.
Also joining the groups.io might be helpful.
https://groups.io/g/FLMesh
73
Matthew KN9U