@d0ugmac1 and @OddErikD 

 

Thanks Doug very much for the extra advice and info!

 

Hi @OddErikD I hope you don't me responding in your thread? I intend to create a thread of my own for Heckington Show soon. I think both your project and mine have similar problems and discussion points.

The info in above posts about positioning/mounting of AP's on masts was very interesting as it was preying on my mind. Plus of course all the weather related aspect.

 

Gig fibre is new in our village. The ISP who provide it will also have a trade stand at the show!

We are getting 1GB business broadband installed by ISP at a house adjacent to the showground fields.

There are no existing major structures for us to attach things to other than what gets put up for the Show, eg lighting rigs, stages, marquees, and various trees and hedges to work around too!

This (first) year we are providing the traders wifi access for free, with no guarantees of throughput. Next year we will fine-tune according to our experience.

 

The house has a cabinet installed for fibre/ONT (optical network terminator), the TP-Link controller, router, TL-SG2008P switch, and the "transmit" CPE710 mounted 6m high on the house beaming 350m over to the second field, to the "receive" CPE710 likely 6-8m high on a marquee pole, down to a TL-SG2008P, then wired/PoE to main uplink EAP650 of Mesh2. Fresnel calculated for this distance line-of-sight over level ground is 2.1m above any obstacles, which there will hopefully be none/few.

 

A 100m ethernet/PoE cable will be laid out into the 1st field to the uplink EAP650 of Mesh1.

 

In theory, each mesh physical location/spread will cover about 125 trade stands, which I would prefer and makes sense to be encompassed by one SSID. I think the EAP650 is rated for 250 client connections per mesh. I've been using the TP-Link heat map tool to get an initial idea about coverage. However, a lot of this is still unknown to me until I start configuring the VLANs, APs, SSID's and done lots of range testing. I'm very keen to see how well the CPE710 bridge performs at 80Mhz - may need to configure tighter rate limits in Mesh2?

 

One of the bigger issues are Apple phones which love to hunt around for 'open' Wifi signals, this generates a lot of noise in the active wifi bands.  My suggestion is to encrypt and hide (no broadcast) your SSID(s).

 

That's good to know. For PCI compliance we will be encrypting, and nothing will be "open". Probably going to use captive portal and issue wifi access "vouchers" to provide the traders with dedicated authorisation. I did wonder about hiding SSID's.

 

Regarding the Omada mesh architecture, be aware that the same 5.8 Wifi channel is used for rx/tx of mesh traffic as well as accepting user traffic from both 5.8 and 2.4 radios.  Given your traffic profiles this is likely less of an issue than most, but my recommendation is still if planning 5xAP mesh (1 root, 4 meshed), it's better to add another AP and make it two sets of (1 root, 2 meshed) and force each set onto a different 5.8 channel.  This also reduces the interference induced when the far left and far right try to talk to the root AP because they can't 'hear' or 'see' each other, they will talk all over each other forcing packet retransmits/loss.  Design your physical layout accordingly.

 

The problem we have is laying Ethernet cables to AP masts where there is going to be lots of people walking and heavy vehicles moving around the showground. We're having to design the layout around locations of generators, and keeping network cabling costs to minimum (almost as costly as the wifi kit!). The total length of the showground is 1km, with wifi over about 700m!

 

I am wary of interference and certainly on 2.4Ghz intend to set Channels per group of 5 APs:- eg. 1, 6, 11, 1, 6. I've yet to measure channel use from neighbouring housing estates. Have you tried using the Omada Auto Optimisation feature, is it good?

 

I thought for 5Ghz it is wiser to choose a least used channel, but I thought it requires the same channel for all APs due to the wireless backhaul? Maybe this is why you're suggesting bridging in the middle of each mesh, to achieve channel x in one half and channel y in other half?

 

Also because we are unclear of what devices (phones/card readers) the traders will be using (new/old), and they are unlikely to be moving about the showground (they should stay at their stand!), we want to minimise use of advanced tech, e.g. WPA3, roaming, beamforming (not important for an open air site??), and such like. Also mindful of reducing transmit power, and setting sensible Rate Limits (thinking about bored traders waiting for custom, downloading videos, posting on FB/Tik-Tok, lol.) certainly until we get a better idea of how well the system performs.

 

We've got 2 months for in-the-field testing before July 27/28th, and there's a lot to do! I started on this project in January '24, lol.

 

Thanks

Jim

Thanks Erik for letting me post here, the next-stage of your harbour project sounds very interesting too. I'll be back with future updates. Best wishes, Jim

  @OddErikD @UKJim 

 

Gents, a word of advice...if trying for a long haul critical backbone line...don't use 2.4 or 5.8G bands and expect reliability.  I'd recommend you jump straight to 60Ghz, something like the Ubiquiti AirFibre line.

You would get better performance if you could organize your mesh as below, ie a string and not hub/spoke.  The issue with the latter, if we look at the LHS group, would be that A could try to talk to root C at the same time as either D or E.  It's better if A only talks to B, and B only talks to A and C, and C only talks to B and  D, and so on.  If you can double up the root AP, then you can run Freq.blue and Freq.red for even better performance.  Of course, this is year 1 and you are really not moving a lot of traffic, so your design will likely work, but if the requirements get heavier, you'll need to up your 'mesh game' :)  With meshing, binary trees are better than trinary.

 

You need to make sure you force (fix) the RF channel, at least for the two Root nodes, don't leave them on Auto, unless you enjoy mass confusion and spaghetti meshing.

 

Hi   @d0ugmac1 

Thanks for your further advice.

I understand about fixing the frequency/channel for the root APs (we don't want Auto changing the foundations if it detects other neighbouring wifi channel conflicts!).

 

I'm unclear about how or what things to configure to achieve what you suggested.
I figured reducing transmit power so as not to overlap or encroach on other AP's in the same mesh is one way.

Do I also need to lock mesh AP's to the specific Root?

I'm confused about the freq blue/red suggestion? How does the root set two channels/freq to serve each side at once?

 

The Controller is quite a beast to get my head around, think my brain is about to explode!

Jim

 

 

  @UKJim 

 

Ok, inline...

I figured reducing transmit power so as not to overlap or encroach on other AP's in the same mesh is one way.

--reducing power for wireless meshed nodes can cause more trouble, its better to address with smart frequency planning

 

Do I also need to lock mesh AP's to the specific Root?

--the controller allows you to set a preferred root AP, I definitely recommend this approach. I show one below with the 'gold star' via Action

 

I'm confused about the freq blue/red suggestion? How does the root set two channels/freq to serve each side at once?

--blue/red is only possible with two root APs at the same point, each with a different mesh freq configured.

 

The Controller is quite a beast to get my head around, think my brain is about to explode!

--you'll know how 'interesting' it will get once you start to see the mesh self organize...hope you don't wind up with 6 APs on one root and 1 on the other :)