General question re. wifi devices disconnecting and reconnecting

[Ian Manning]

For many years I've had a problem with wifi devices dropping off the network and being unable to reconnect without me rebooting either the router or the mesh network. I have 40-50 devices connected to the wireless network at any one time. Some of the devices in question have static IP addresses reserved on the router, others don't. This doesn't seem to affect whether or not they suffer from this problem.
My router is a Netgear R7800 running DD-WRT (it also acts as my DHCP server). My mesh network is a BT Whole Home Wifi with 4 discs. They're now all connected via ethernet, so I guess it's technically not a mesh network any more, but a group of managed access points.
My questions are:
1) what are the most common reasons for devices dropping off the network and not being able to reconnect automatically?
2) why is it sometimes possible to fix the problem by restarting the mesh network/access points (and not the router)? (since my understanding was that the router was in control of allocating IP addresses)
3) why will devices often not reconnect to the access point which is nearest to them? For example my video doorbell often connects to an access point at the back of the house, even though there is an access point 6 feet away from it, on the floor above.
I'd like to upgrade my way out of the current situation if that's possible, but first need to try to understand the possible root causes.

 

[OzarkEdge]

For many years I've had a problem with wifi devices dropping off the network and being unable to reconnect without me rebooting either the router or the mesh network. I have 40-50 devices connected to the wireless network at any one time. Some of the devices in question have static IP addresses reserved on the router, others don't. This doesn't seem to affect whether or not they suffer from this problem.
My router is a Netgear R7800 running DD-WRT (it also acts as my DHCP server). My mesh network is a BT Whole Home Wifi with 4 discs. They're now all connected via ethernet, so I guess it's technically not a mesh network any more, but a group of managed access points.

How can you manage networks and permit this for years???

My questions are:
1) what are the most common reasons for devices dropping off the network and not being able to reconnect automatically?

Which network? All of them? What does drop off mean exactly?

2) why is it sometimes possible to fix the problem by restarting the mesh network/access points (and not the router)? (since my understanding was that the router was in control of allocating IP addresses)

Maybe drop off means your wire backhauls/switch are faulty. Cut it down to one router, one new/trusted/temp wire direct, and one AP and test.

3) why will devices often not reconnect to the access point which is nearest to them? For example my video doorbell often connects to an access point at the back of the house, even though there is an access point 6 feet away from it, on the floor above.

Too much WiFi despite the solid walls?
Localized radio/EMI interference/obstacles?
Crappy IoT firmware?
Faulty wired backhaul corrupting the AP operation?
Failing router LAN ports?
AP directly over the client location?
Faulty AP signal/radio?
A wired/LAN device failing and pulling down the local electronics (test with everything disconnected)?
A floating power ground/Earth connection on that AP circuit/at the building demarc, corrupting AP/network electronics?

One rule I use is that 99.99% of network trouble is cabling... to not overlook this fundamental source of trouble. Maybe the cat has been biting your cables... etc.

You have to narrow down the problem, find cause and affect, swap pieces in and out to test. If you don't like the network and want to upgrade, it may be more practical to get on with developing your new network... build it out one step at a time to isolate trouble before hiding it with too much gear; when ready, turn it ON and retire the old gear and its trouble... to save time and trouble. Keep it simple... it will get complicated on its own as you age and become less interested and able to manage it... and don't let your users suffer for years.

OE

 

[coxhaus]

For many years I've had a problem with wifi devices dropping off the network and being unable to reconnect without me rebooting either the router or the mesh network. I have 40-50 devices connected to the wireless network at any one time. Some of the devices in question have static IP addresses reserved on the router, others don't. This doesn't seem to affect whether or not they suffer from this problem.
My router is a Netgear R7800 running DD-WRT (it also acts as my DHCP server). My mesh network is a BT Whole Home Wifi with 4 discs. They're now all connected via ethernet, so I guess it's technically not a mesh network any more, but a group of managed access points.
My questions are:
1) what are the most common reasons for devices dropping off the network and not being able to reconnect automatically?
2) why is it sometimes possible to fix the problem by restarting the mesh network/access points (and not the router)? (since my understanding was that the router was in control of allocating IP addresses)
3) why will devices often not reconnect to the access point which is nearest to them? For example my video doorbell often connects to an access point at the back of the house, even though there is an access point 6 feet away from it, on the floor above.
I'd like to upgrade my way out of the current situation if that's possible, but first need to try to understand the possible root causes.

My guess is if you are running all the APs on the same channel then it would be a mesh system regardless of layer 1, wire or wireless, as mesh is a redundant system.

Have you tried using different radio channels? In which case it would not be mesh as you are not backing up the same channel. It runs better to use multiple channels with higher through put.

 

[Ian Manning]

I've tried a different channel, having used a wifi analyser on my phone to avoid the most congested ones. By "dropping off", I mean devices losing their connection to the wireless network and hence the LAN.

I suppose I can understand interference, cabling and hardware causing potential problems, but what I don't understand is why simply restarting the mesh network (rather than the router) can immediately restore a device to the wireless network??

 

[coxhaus]

I've tried a different channel, having used a wifi analyser on my phone to avoid the most congested ones. By "dropping off", I mean devices losing their connection to the wireless network and hence the LAN.

I am talking about using different channels for each AP.

You can exceed the number of devices supported per AP. Some companies don't publish specs for the number of supported devices. You need both the number of connected devices and the number of active devices. Consumer gear tends to be low.

 

[Tech9]

having used a wifi analyser on my phone to avoid the most congested ones

How do you know which ones are congested when Wi-Fi Analyzer doesn't show channel utilization? This is a common mistake. All you see there is near APs tuned to specific channels. Active or not - you don't know. Routers on Auto around see a channel with high bandwidth available and jump to it. You see many routers there and hurt yourself choosing the channel with the most noise (for example) no one is using for a good reason.

 

[tgl]

You can exceed the number of devices supported per AP. Some companies don't publish specs for the number of supported devices. You need both the number of connected devices and the number of active devices. Consumer gear tends to be low.

FWIW, my gut feeling is that this is probably the core problem. 50 clients is a lot by consumer-gear standards. I have no experience with BT Whole Home Wifi, but googling finds that there are several different models (which one have you got?) and for none of them do the specs mention the max client count.

 

[tgl]

1) what are the most common reasons for devices dropping off the network and not being able to reconnect automatically?

@coxhaus fingered a likely cause, I think. The fact that this works:

2) why is it sometimes possible to fix the problem by restarting the mesh network/access points

indicates that it's a wifi saturation problem and not a network problem.

3) why will devices often not reconnect to the access point which is nearest to them?

That's a very common symptom of a badly-tuned multi-AP setup. Often, clients will connect to the first signal they scan that is "strong enough" according to some built-in threshold rules. A typical recipe for improving that is to move the APs further apart and/or reduce their transmit power, so that there's not so much overlap in their coverage. I wonder though if these BT units let you set the Tx power at all; the reviews I found online didn't seem impressed with their configurability.

I would suggest reducing the number of APs, except that if the theory that you're exceeding the max client count is correct, that might make things worse not better. It's still worth a try though, if only to gather evidence. @OzarkEdge 's advice about systematically changing things to narrow down the problem is 100% on point.

 

[Ian Manning]

Thanks for the feedback. I have the original version of the BT Whole Home Wifi discs. There is no indication of a limit to the number of clients per access point. There is also no per-disc configuration of channel number or signal strength. The only per-disc configuration option is the name of the disc.

BT support haven't been much help. They advised me to disconnect all ethernet from the non-primary discs and see if the problem persisted (it did) - I guess this was intended to eliminate any ethernet cabling issues as a possible cause?

In the most recent occurrence of the problem (yesterday) I had a TP-Link Tapo L530 Smart Bulb which could not reconnect to the wifi network. The access point nearest the bulb (18 feet away) only had 4 other clients connected at the time. Does this suggest that the root cause is not the limit on clients per AP? I tried soft resetting the smart bulb, restarting the mesh network, then restarting the router, but it still will not connect.

 

[tgl]

I continue to think that the root cause is "too many clients for this not-very-high-end consumer wifi gear". How that's manifesting exactly is not clear. But given the lack of visibility into the APs' configuration or operation, I think trying to chase it down is likely a fool's errand. You should be thinking about replacing all the APs with something more modern.
You've not given us any info about the size or layout of your place (details like what the interior walls are made of matter, too), so it's hard to offer concrete advice about what to replace 'em with.

 

[Ian Manning]

The house is a 3-storey brick-built townhouse (1970s build) with insulated cavity walls. The router and primary mesh disk is in the study - middle of the ground floor. The 2nd disc is at the back of the ground floor, mainly to provide coverage for the garden, The 3rd disc is on the 1st floor, on the window sill at the front of the house - to provide coverage for the video doorbell and security camera (about 6 feet below it). The last disc is on the 2nd floor, in the main bedroom at the back of the house.

Any suggestions for "something more modern"? Maybe Deco M5?

 

[tgl]

You didn't mention square footage, but those AP placements sound sane as long as the place isn't too large. And certainly, if you already have ethernet drops at those spots, that'd be good reason not to move them without great need.

As for what to use instead ... it depends a lot on what kind of management experience you want, and whether you are looking to replace your router or only the APs. If you are happy running DD-WRT then I suspect you might be the kind of person who would like more control over the APs than what you have with the BT discs or are likely to get with gear like the Decos. If I'm wrong about that and you just want a drop-in replacement for the discs, you probably won't go wrong with Decos. (The M5 is pretty far from bleeding-edge gear though. Are you looking to acquire WiFi 6 or newer clients any time soon? If so you might want a newer model.)

If you do want more control then I think you'd do well to look at SMB (small/medium business) gear, such as from Cisco, Omada (also TP-Link, but a different product line), Ubiquiti, or Zyxel. This kind of product line will include APs that are just APs, not all-in-one router plus AP which is what consumer devices inevitably are. The hardware cost savings from that partially offset the fact that business-oriented gear tends to cost more, so that you aren't necessarily spending a lot more. As an example, the best price I'm seeing on Amazon for a Deco M5 3-pack is $130, while the roughly comparable specs-wise Zyxel WAC500H goes for $70 apiece. Both of those are kind of bottom-of-the-line, to be clear, and if I were buying something today I'd be looking for at least WiFi 6 (AX) hardware.

 

[Ian Manning]

The ground floor is approx 650 sq feet, and the upper two floors are approx 450 sq feet.

I'd like a little more control over the AP's - ideally to be able to:
- direct/force specific clients to connect to a specific AP (I believe that this is possible with some of the Deco models?)
- restart individual AP's from a central console.
I'd also like roaming between AP's, for devices that move from room to room.

My instinct is to try to keep the router separate from the AP's if possible, simply to give me more flexibility to replace either one independently in the future, rather than being tied into a particular ecosystem. I really only run DD-WRT for the QoS - to prevent large downloads or P2P from interfering with general internet access and streaming. I have alot of port forwarding and static IP addresses defined on the router, but other than that I'm more interested in monitoring (in the event of problems) than sophisticated configuration options. I had a quick look at the Omada range, and the vast majority of features really don't sound relevant to me. The phones and laptop are currently the only wifi 6 devices in the house, but I guess other wifi 6 devices may appear slowly over time. I do have alot of connected devices (40-50) - cameras, home automation gear, etc.

My gut instinct is that SMB gear may be overkill for me - maybe 80-90% of the features wouldn't be relevant to me. I suppose I'm looking for a step up in terms of control of AP's but not a sea change in terms of configuration and management. I must admit I am struggling to make sense of the vast Deco range, and identifying which would be most appropriate for me. I guess the X series with Wifi 6 would offer future-proofing, but there are at least 3 models in that range and I don't know which offers the best match for me. I'm not very familiar with Zyxel gear, but the WAC500H seems to retail for around £160 apiece here in the UK.

 

[degrub]

i had the same feeling and similar issues about 10-15 years ago. Switched over to SMB gear ( CISCO in my case RV325 and WAP371 APs). LAN and WAN stable as a rock from day one. Didn't need to use 90 % of the control offered, but everything i did need - basic access control, VLANs, firewall rules, etc. just works. CISCO has adjusted their product offering, so all i would suggest is the APs (150) and a Layer 3 switch plus whatever VLAN aware router. TP-Link Omada would be my other choice with an OC-200 controller.

For such a small volume to cover, you may be better off running just 5GHz on your APs and turn off the 2.4GHz radios. Your interference is likely from neighbors blasting just as hard as your APs. 5GHz will penetrate one sheetrock wall with enough signal strength usually and is slightly less crowded.

i have a roughly square two story, each floor 1500 sqr ft . i did basic wireless strength survey to rough in where i needed APs and what signal strength. i use 5Ghz only, and reduced power on most radios. i use the walls to further limit to avoid significant overlaps. i have both wall mounted and ceiling/table mounted APs. i don't get wifi call dropouts when roaming the house. i have 2 APs upstairs and 2 downstairs. The ones towards the back are oriented horizontally and opposite - one ceiling 1st floor, one on table 2nd floor to take advantage of the directionality of the radios. The other two are wall mounted near the front wall of the house to cover 1st and second floor rooms where there are too many walls and foil covered ducts in the way. i use coax (moca 2) as my backbone to link all devices on ethernet.

If you need 2.4 GHz band for some devices, you will want to turn the signal strength down as much as possible and locate the AP near them if possible. Maybe only run one AP with 2.4 GHz ( which eliminates using the CISCO single point setup i think, although you might be able to 0 the power out - haven't tried for my case) if possible.

It sounds like you have an elongated foot print roughly 15 x 30 ft for the two upper floors. i assume the 650 sqr ft includes the garden area. To cover the garden, you may want to use a low power outdoor rated AP under the roof eave for more flexibilty inside. You could use the same approach for the front door devices.

Just some ideas. Switch to the SMB gear if you can. The DECOs may work well enough. You'll have to experiment.