Wifi 2.4GHz signal dropout frequently

[kohjb]

I am using an AX11000 meshed with 3 ZenWifi AX Minis. I'm noticing very intermittent Wifi connections, especially on 2.4GHz. Using Wifi Analyzer, I've observed that the 2.4GHz singal seems to routinely drop out once every 3-5 seconds (see graph below). Can anyone help to explain why this happens from a firmware perspective? And/or provide a solution to keep the 2.4GHz signal more stable?

Thanks in advance!
JB.

 

[kohjb]

Just as a follow up - the above graph of the signal could be just a red herring I'm afraid. That graph was obtained when I was on a 5G network, looking at the performance of my 2.4G signal. When I connected to the 2.4G network (the very same one I wanted to measure), the signal had NO dropouts at all! So it could well be the phone/tablet that I was using to measure the signals that is unreliable :-( Oh well, back to the drawing board.

Best regards,
JB.

 

[tgl]

I am using an AX11000 meshed with 3 ZenWifi AX Minis.

That's a lot of wifi. How large is the area you are trying to cover? It could be that the APs are too close together and the instability traces to clients hopping from one to another constantly.

 

[kohjb]

I'm in a 3 storey home, and the house has concrete walls and brick walls internally. So I have 2 on the ground floor, one in the back room where the Wifi just doesn't get to, and 2 on each floor. All the nodes are connected via wired LAN.

Yes, I wanted to understand how to identify (is there a tool that can read a metric) of the collisions occuring. I did try to space the nodes apart so that they don't overlap too much with each other. If there's an objective way of measuring and/or understanding what's going on, that would be great.

As mentioned in my follow up post, the dropouts don't seem to be happening IF the phone I'm using to measure is on the same network. However, I'm confused as to knowing what is real (e.g. are the dropouts because it's the phone that is turning off or throttling down the wifi scans, or is it that when the phone is on the same network, it is artifically showing that the signal is stable and not dropping out?).

Thanks!
JB

 

[OzarkEdge]

I did try to space the nodes apart so that they don't overlap too much with each other. If there's an objective way of measuring and/or understanding what's going on, that would be great.

Maybe this can help with node placement:

o enable wireless backhauls, you can observe their connection details in the router Wireless Log when not using wired backhaul

o adjust the wireless backhaul distance/path for 5.0 connection RSSI >-66dBm (my number... feel free to go more negative within reason, say -72dBm); router-node wireless connections should be best-case given their radios and antennas vs lesser client hardware

o disable/de-prioritize wireless backhauls to resume using wired backhaul

OE

 

[kohjb]

Maybe this can help with node placement:

o enable wireless backhauls, you can observe their connection details in the router Wireless Log when not using wired backhaul

o adjust the wireless backhaul distance/path for 5.0 connection RSSI >-66dBm (my number... feel free to go more negative within reason, say -72dBm); router-node wireless connections should be best-case given their radios and antennas vs lesser client hardware

o disable/de-prioritize wireless backhauls to resume using wired backhaul

OE

Thanks for this great suggestion - very logical now that you mention it! Will try.

A few questions before I do that though.....My current setting for the nodes for Backhaul Connection Priority is set to Auto. I can choose 5GHz Wifi first, or 1G WAN first. I assumed that Auto would always default to WAN first? Or is this incorrect and I should set it to WAN. I did notice in the Wireless Log that there are entries there, but no mention of distance/path - so can I assume backhaul is via WAN?

Also, there's Preferred WIFI Uplink AP - which can be set to Auto, the main router, or the nodes. It was on Auto, but I've set it to the Router. Is this setting important if I'm using a Wired backhaul?

Thanks so much again for the help and suggestions!
JB

 

[OzarkEdge]

A few questions before I do that though.....My current setting for the nodes for Backhaul Connection Priority is set to Auto. I can choose 5GHz Wifi first, or 1G WAN first. I assumed that Auto would always default to WAN first? Or is this incorrect and I should set it to WAN. I did notice in the Wireless Log that there are entries there, but no mention of distance/path - so can I assume backhaul is via WAN?

Assuming Ethernet Backhaul Mode is disabled to permit wireless backhauls, leave it on Auto and disconnect the wired backhaul... the wireless backhauls should appear in the log if not already running. Then you can read their connection RSSI in the log (no path/distance data there).

Also, there's Preferred WIFI Uplink AP - which can be set to Auto, the main router, or the nodes. It was on Auto, but I've set it to the Router. Is this setting important if I'm using a Wired backhaul?

I have not used it (one node here). I would prefer to not use it, to not force things, by striving for a star topology that has all wireless backhauls connecting directly back to the router.

Once the physical layout is determined radio-wise, I prefer to enable Ethernet Backhaul Mode to disable ALL wireless backhauls (all WiFi for client use only and no failover to wireless backhauls) and rely only on the wired backhaul... it's plenty reliable by itself for this class of networking, imo. In this case, node Backhaul Connection Priority can be set to WAN/Ethernet only, or Auto.

Auto is suppose to default to the best backhaul AiMesh detects... it might choose wireless over wired if the wireless is 'better'. Disabling the wireless backhauls avoids this uncertainty and any sporadic switching back and forth between wired and wireless backhauls in our amateur installations and the ensuing disruption to the user experience.

OE

 

[kohjb]

Assuming Ethernet Backhaul Mode is disabled to permit wireless backhauls, leave it on Auto and disconnect the wired backhaul... the wireless backhauls should appear in the log if not already running. Then you can read their connection RSSI in the log (no path/distance data there).



I have not used it (one node here). I would prefer to not use it, to not force things, by striving for a star topology that has all wireless backhauls connecting directly back to the router.

Once the physical layout is determined radio-wise, I prefer to enable Ethernet Backhaul Mode to disable ALL wireless backhauls (all WiFi for client use only and no failover to wireless backhauls) and rely only on the wired backhaul... it's plenty reliable by itself for this class of networking, imo. In this case, node Backhaul Connection Priority can be set to WAN/Ethernet only, or Auto.

Auto is suppose to default to the best backhaul AiMesh detects... it might choose wireless over wired if the wireless is 'better'. Disabling the wireless backhauls avoids this uncertainty and any sporadic switching back and forth between wired and wireless backhauls in our amateur installations and the ensuing disruption to the user experience.

OE

Thanks OE for the reply! Much appreciated!

 

[drinkingbird]

I'm in a 3 storey home, and the house has concrete walls and brick walls internally. So I have 2 on the ground floor, one in the back room where the Wifi just doesn't get to, and 2 on each floor. All the nodes are connected via wired LAN.

Yes, I wanted to understand how to identify (is there a tool that can read a metric) of the collisions occuring. I did try to space the nodes apart so that they don't overlap too much with each other. If there's an objective way of measuring and/or understanding what's going on, that would be great.

As mentioned in my follow up post, the dropouts don't seem to be happening IF the phone I'm using to measure is on the same network. However, I'm confused as to knowing what is real (e.g. are the dropouts because it's the phone that is turning off or throttling down the wifi scans, or is it that when the phone is on the same network, it is artifically showing that the signal is stable and not dropping out?).

Thanks!
JB

The network you are not connected to is not going to show the proper strength and will vary like that. Power saving in action. If you want to monitor strength/reliability you need to be connected to that network.

 

[kohjb]

The network you are not connected to is not going to show the proper strength and will vary like that. Power saving in action. If you want to monitor strength/reliability you need to be connected to that network.

I see. Thanks for the explanation. I was wondering how wireless scanning works in the Android phone, so this provides a glimpse.

Cheers!
JB

 

[drinkingbird]

I see. Thanks for the explanation. I was wondering how wireless scanning works in the Android phone, so this provides a glimpse.

Cheers!
JB

You can defeat a lot of the power saving and scan throttling in developer options but you won't be able to get rid of it completely, plus you'll sacrifice a lot of battery life. For some, disabling scan throttling is needed (i.e. you roam between APs a lot when on wifi calling etc) but if you don't need to, just wastes battery.

 

[kohjb]

You can defeat a lot of the power saving and scan throttling in developer options but you won't be able to get rid of it completely, plus you'll sacrifice a lot of battery life. For some, disabling scan throttling is needed (i.e. you roam between APs a lot when on wifi calling etc) but if you don't need to, just wastes battery.

I've been having it on, and honestly haven't noticed any significant drop in battery. But then again, my phone and battery are still relatively new so perhaps has lots of juice

I did turn it off already all the same, since one of my banking app complains if developer mode is left on :-(

Cheers!
JB

 

[drinkingbird]

I've been having it on, and honestly haven't noticed any significant drop in battery. But then again, my phone and battery are still relatively new so perhaps has lots of juice

I did turn it off already all the same, since one of my banking app complains if developer mode is left on :-(

Cheers!
JB

It's been this way even on PCs for a long time (even back when inSSIDer was called NetStumbler), even before all the power saving stuff, the networks you are not connected to are not going to provide real time, accurate information, since your device is just periodically seeing that network, rather than being actively connected to it with consistency. For seeing the environment, it is fine, but for second by second stats, you really need to be actively joined to the network. But Android (and possibly iOS, no experience with that) is very aggressive with power saving, even when you disable the scan throttling.

 

[kohjb]

It's been this way even on PCs for a long time (even back when inSSIDer was called NetStumbler), even before all the power saving stuff, the networks you are not connected to are not going to provide real time, accurate information, since your device is just periodically seeing that network, rather than being actively connected to it with consistency. For seeing the environment, it is fine, but for second by second stats, you really need to be actively joined to the network. But Android (and possibly iOS, no experience with that) is very aggressive with power saving, even when you disable the scan throttling.

Yes, understand. I guess my reference is that I've been using this app called WiFi Analyzer (which basically hasn't changed or been updated since 2018) and have always used the timeline graph on my android phones, and it never depicted the dropouts or unstability. The dropouts appeared only relatively recently (say last 2-3 years?). So it could be the app faking it, or the new android versions doing lots of throttling of the scans, that's causing the drops in signal. In any case, I'm thinking of purchasing a standalone device (so my surveys are independent of android's updates) like this WIFI Signal Scanner , wonder if it'll be any good?

Thanks and regards!
JB