How many wifi devices is too many?

[thiggins]

• ~15 wifi camera devices
• ~20 pcs/phones/laptops
• ~10 IOT
• Around 20 wifi clients connected to 1 router and another 10 to the aimesh node

Going back to the top, you're saying you have 45 devices, but provide connection breakdown for only 30. Where are those other 15 devices?
If you can fill out this table, for with and without guest loadings, it can help diagnosis. If you can provide the Wi-Fi flavor the devices use, that will really help understand potential airtime use.

Root Node
2.4 GHz
5 GHz

Leaf Node
2.4 GHz
5 GHz

It's with the root node. I actually never tried pinging the leaf nor have I checked whether they are connected to leaf or root node. Hmm I guess I should find that out.

Also, when you ping, it's important to know which band the client you're pinging from is connected to.

I've been wanting to go for wifi6 as my neighbors have started to use the 5G network and they seem to have the google mesh.

You may be confusing Wi-Fi 6 and 6E. Only 6E supports 6 GHz, but then you need 6 GHz devices to get away from 5 GHz.

 

[banz]

Going back to the top, you're saying you have 45 devices, but provide connection breakdown for only 30. Where are those other 15 devices?
If you can fill out this table, for with and without guest loadings, it can help diagnosis. If you can provide the Wi-Fi flavor the devices use, that will really help understand potential airtime use.

Root Node
2.4 GHz
5 GHz

Leaf Node
2.4 GHz
5 GHz

Also, when you ping, it's important to know which band the client you're pinging from is connected to.


You may be confusing Wi-Fi 6 and 6E. Only 6E supports 6 GHz, but then you need 6 GHz devices to get away from 5 GHz.

My other 15 devices are just random things. Mostly just smart devices, bulbs, light switches, power switch, google home, amazon echo, ovens, vacuums, raspberry pi. They operate all on 2.4G and I try and keep my 5G solely for devices that need high bandwidth such as tablets, mobile phones and laptops. Currently I only see around ten 5G devices.

Most of the 2.4G devices just drops off the Clients "view list" when they are not in used. Plus the yazfi guest network devices only appears on System Log- Wireless logs. This does not show me if it's connected to root or leaf node. I guess I'll need to run something like airmon-ng to see which bssid it is connected to.

Yes I was confusing wifi 6 with 6Ghz.

 

[banz]

This is the 2.4G band and I'm using auto. Currently it's at channel 10.

However the guest are all connected to the 5G network. My thoughts were the 2.4G was overloading the router so it was causing problems for 5G band. After reading so many advices/suggestions on here, it seems 2.4G clients should not affect 5G in any way.

I've also removed the devices that are plugged into the main router usb 3.0

 

[banz]

I only have one neighbor running the same channel as me for 5G.

 

[L&LD]

With an ISP upload speed of only 10Mbps, you may need to enable QoS on your network to share that very low bandwidth fairly.

Yes, any diagram you can provide will be welcome. Hard to visualize your setup without it.

And running 'Auto' for the Contol Channels is probably the biggest issue here.

On the 5GHz band, use another Control Channel (other than what your immediate neighbor is using).

 

[banz]

I've attached the floor plan. Bear in mind that the router/leaf node is located at those location because there are existing cable infrastructure and it is a rental home. So I cannot start running cables/drilling holes through the walls.

We are mostly all at Room 2 on the 1st floor as it's the living room. Room 2+ 3 does not have a wall. Room signal isn't great at room 4 and most clients connect to the router on the second floor if we go into this room.

As suggested, I've changed the control channels to static. I looked for the range with the weakest neighbors SSID and moved my channel to it. I actually see an improvement straight away with my 5G band. My phone now gets full 5 bars signal anywhere in the house. Previously it might dropped 1-2 bars if I go to the edge of the house.

 

[alba666]

In my experience, routers/nodes hugging walls degrades performance. Ideally, you want clear line of sight > door > inside wall > outside wall between your WiFi AP’s and devices.
In my case, running coax behind a couch to get my main router off the wall and close to home center line removed the need for an AIMesh node.

 

[heysoundude]

I'm using ac68p and ac86u (primary). I first started with the ac86u only. I don't recall seeing as much problem with just 1 primary router. However it has many dead spots in the house and the signal is too weak in some areas hence why I started adding more nodes.

How willing are you to experiment?
It's a random "hail mary pass" or "shot in the dark" type of experiment, but :

Router temperatures

just ordered a new dedicated firewall hardware If you are planning to re-use the AC86U as Access Point, Asuswrt doesn't have VLAN's support and you'll lose Guest Network option. It is actually available in AP Mode GUI, but it doesn't separate Guest Network connected devices from your main network.

www.snbforums.com

maybe jumbo frames might help the backhaul connection between the main router (AC86) and mesh node (ac68), if the cable between them meets Cat5e or higher. If it does, maybe any wireless bottleneck might open up. And then if it does, add the mesh nodes you've removed back into the mix, but reduce each of their Tx Power adjustment (if that's possible: Advanced Settings - Wireless - Professional) from "Performance" to "Good" or "Balance" or even "Fair".

My supposition is in two parts: that many of your wireless clients are having a difficult time deciding which Access Point to connect to, and they may often be blasting radio frequency energy out to reach the one they might think (incorrectly, possibly) is the best - all that energy from some clients might be interfering with the rest. putting more APs as you did, all at full power, seems to have only made matters worse, which is possibly why you removed them. Ignoring the labels in this, consider each of the circles is the wireless bubble of each WAP/node:


my suggestion: make more "bubbles," then make them "smaller" so they don't overlap as much, and perhaps it might be easier for the clients to choose which to associate with, possibly opening up the airwaves for all of them. Then, if enabling jumbo frames makes even more room in LAN packets (the second part of my supposition, which you can test easily first), you might just smooth out the issues you're having.

this presumably would get the clients closer to fewer access points without sacrificing coverage, and transfers information to the router for use in the LAN more efficiently/effectively, and then out the WAN connection (if applicable).
Getting QoS happening would probably help as well for both uploads as well as downloads

If this makes enough sense for you to try it, and you get good/better results, I hope you'll let us know. I have faith in the technology, and that it is possible to configure it in ways to make life easier/better for everyone using it

 

[thiggins]

@heysoundude Jumbo frames have not been needed for years now due to improved processing by newer PHYs.

Co-channel interference (too much overlap between APs) would be a problem if the OP said that devices were not roaming properly or disconnecting/reconnecting a lot. I don't think he said that.

 

[heysoundude]

<shrug> I did make the disclaimer that the ideas I suggested were experimental, throwing a dart over the shoulder hoping for luck to hit a bullseye. If the OP is making several posts/threads, maybe they feel trying anything might be worth a shot - clearly they're looking for several perspectives/options/viewpoints. Only they can tell us what works in their environment, if one of the things here manages to sort out the troubles.

PHY improvements may generally be the case, but did the router design/software folk incorporate them or take them into account? <shrug> can't say, and trying it could make things worse...but it might make things better for some silly reason we're not factoring in. Reverse engineering (or whatever you may want to refer to this as) may reveal a deviation from standard practices for whatever unknown reason. I'm trying to think broadly and give benefit of doubt.

RF can be funny, and that's why anybody who has mission-critical requirements of their network tends to prefer wired backhaul connections rather than relying on potentially funny wireless. especially in non-perfect environmental conditions. plus, there is a weird assumption that if there's wifi, it'll work well/perfectly. The science and engineering may be sound and the technology properly deployed, but...right?