Hi, first time posting here. I have some WiFi experience (not MikroTik, AP's and p2p's) and some limited MikroTik router experience, so with that in mind:

I'm looking to provide, for 11 cabins in a rough circle, WiFi via 2.4GHz, but I need to connect them all together via WiFi too. I'm thinking 2 antennae, one inside, one out. I understand this to be a WDS for the outside stuff and multiple cabins have ethernet to them which I'll need to use one of to get the WDS wired.

After looking at the RB products, I thought an RB411AR coupled with an 2.4 omni antenna, plus an R11e-5HnD, coax to an outside 5 omni. Seems like the 2.4 power is good for interior use and the 5 has enough power to link the 10-100 feet between cabins. If there is a better alternative, what would that be? Most devices seem to be single frequency.

In the future, when the wiring fails to these units, I'll have to provide VoIP, so again the RB411AR looks like it can do what I need with the ethernet port, which is tag vlan 2, required for the phone, and leave remaining traffic untagged. The vlan extension can be stripped over the WDS, I can manipulate the tag in/out of the infrastructure switch.

Q1: Can WDS provide QoS to ensure the VoIP traffic is prioritized as it hops around?
Q2: Can I add a user SSID to the WDS on 5GHz to provide outside-WiFi and maintain the VoIP QoS integrity?
Q3: I notice the R11e-5HnD has two antennae. Can I put two directional antenna on those if I have limited LOS?

All suggestions are gratefully welcome.
Thanks -- lee

It mainly depends on what you want to achieve. Maybe you can share the expected performance numbers. If you do not mind that all devices including your APs and all the clients will share the same frequency, then wds would be the possible solution. But I would not recommend it.

I would not recommend also to run coax cable for long distances as it will be just another source of problems.

Instead of that I would suggest following:

Make one central point where connectivity is attached to with with 5ghz ap (for example SXT SA5 if it would be on the side from the clients or omnitik if it will be in the middle of the clients). Use some outdoor client devices (for example SXT5Lite) to connect cabins to the central point individually. Inside the cabin use some cheap ap, for example hAP Lite for 2,4Ghz AP and internal wire connections. With such configuration you can use vlans with no restrictions.

If some cabins are close to each other, you will not probably need separate client+ap in each cabin, so you can spare some money here...

Anyway, running wires everywhere is much better that thinking about wifi.

Sorry I wasn't too clear, but the WDS I was describing would be 5GHz outside the cabins, and the users would connect with 2.4GHz inside. However I realized that if I only have a single ethernet into the circle of 5GHz WDS that may not work so well, especially for VoIP because it may end up having to go through 10 units to get to the wire.

The coax is to run an antenna from the device in the cabin to the antenna outside the cabin, so it's not a long run, just a pigtail.

So what you describe is precisely what I'm after operationally, the difference being instead of WDS, its a multi p2p. Therefore the QoS is greatly simplified.

Running cable is not an option, it's too expensive in terms of we have to cut through a road 9 times, there's direct buried cables that are unmarked, so we risk cutting through stuff, septic tanks in the way and a lot of the distances are all near the 300ft mark so we can't go around things. I have fiber to one cabin simply because it was 350ft to get around all the junk.

Let me dig up a map and see what the sight-lines look like.

Thanks a lot -- lee

ok, lets show the map.

The client-ap structure is more effective because the ap can be the only point that allows the clients to talk. What is more, as there will not be necessity for transferring data from one client to other client so much, you can simply isolate them and they will not waste time talking together uselessy. On the other side, when using wds mesh, everyone talks to everyone so there is a risk that two radios decide to talk to other two radios at once so they will interfere to each other. Then they will stop, wait and try again. Can be robust but not effective.

WDS is good in case you need to use the same radio as client and ap or when the structure could change very often and you need to have living structure that somehow finds always some way out. Your case looks statically, and all cabins will be pure clients. Moreover, when meshing, you would use omnidirectional antennas that have low gain and catching everything from around so the signal-to-noise ratio is not so much good like if you use panel or dish antenna on the client that is able to attenuate signals from all around and clearly receive only the signal from desired direction (from the AP). It is good for ap also, that the clients are directional, because they can deliver much stronger signal to the AP, so even it has omnidirectional antenna, the signal from wanted radios is way much higher than the signal from unwanted interfering radios somewhere more far around aiming at other directions. Due to this you can use high modulations, transfer more data more quickly and have more stable network.

You can also work with transmitting power of the ap and clients and finetune it so that all clients will look at the ap like they are the same powerful (compensate the distance by transmitting power). If they will receive the signal from ap clearly, they will keep the high data rates all times. At these distances you will have a plenty of power, so you will need to lower the transmitting amplification to very low values. It also saves the electricity pretty much.

So, cobination of omnitik as ap (if in the middle or some panel type ap if on the edge) and sxt as clients (and hAP as internal local accesspoints/connection points) would be very easy to deploy, to manage and quite cost efficient.

What connectivity you want to provide to individual cabin and what connectivity you have available in total at that place?

And you will not need to play with any pigtails. Try always to avoid them because many of originally unexplainable problems are caused by them.

100mb/s coming into the central area via fiber, into 24 port managed switch
3mb/s down, 3mb/s up internet available.
128kb/s down, 128kb/s up VoIP per cabin (potentially 1.5mb/s total, worst case)

So at most we're looking at 3mb/s + 1.5mb/s in both direction simultaneously through the central radio and for each cabin either 3mb/s+128kb/s if only one cabin uses internet or 0.3mb/s+128kb/s if internet equally distributed.

So this map does not have the trees on it and everything is based from me at ground level:

1 -- No visibility to central point, perfect LOS to 2
2 -- Possible LOS to central point, perfect LOS to 4
3 -- No visibility to central point, only 4ft away from 4, can run wire or p2p radio
4 -- Perfect LOS
5 -- If the beam is tight, perfect LOS (between two trees), otherwise LOS to 4
6 -- LOS to central
7 -- LOS to central
8 -- LOS to central
9 -- LOS to 8
10 -- LOS to 9, maybe count make to central point, but doubtful
11 -- LOS to 9, tight beam could make to central point

Red bold -- good LOS
Red dim -- possible LOS



Here's #5 and the spot it would have to hit on the central building. Is the 5GHz beam tight enough to do this.



So I think I may need a simple series of p2p to handle the 8/9/10/11 case if 8 is the only place I can pickup a signal. Similarly with 2, if I cannot get a signal from central, I could p2p it from 4, then p2p it to 1.
And the QoS for the VoIP remains sane with this setup, albeit lots of devices to setup on.