that's what I meant by full power, full allowable.
No, I don't think they even set it to the maximum allowable. High power increases the temperature of the router which would affect the lifetime of the router. But anyway, I guess he will have to check the coverage of each AP using a WiFi scanner and work on reducing overlaps.
technically the min(max allowable, max possible)....
As here, for example, see above in screenshot where under "Transmit Power:" it show "low, middle, high" and high is selected... I've never seen a factory firmware that does anything other than selecting max out of the box... perhaps the higher end ciscos have some useful "auto" function, but even "auto" probably just means "max" for most firmwares.
Cutting power is fine, that's the right way to do it and SOP when setting up any sort of multi-cell radio system.
Forcefully deauthing or refusing to give a DHCP lease as has been discussed here is a whole other story.
And I am fairly certain both Cisco and Ubiquiti's UniFi enterprise APs do the sort of automatic power control mentioned by another user. I know I've seen UBNT reference it in marketing material.

enterprise APs
sure, Cisco and ubnt are likely to do better.
forcefully deauthing on received signal strength reportedly works well, APs are often much louder than the phones they serve so they are able to figure out when to drop a client better than the client can figure out to roam... but DHCP is a big hack, I agree it isn't the right approach
- Limit by DHCP
- Limit by Max Associations in Wireless.
Is the end effect different on client?
Will the client try to reconnect to a different AP?
Association happens long before DHCP, and there is a matching de-association event. Of those two options, I would prefer association-based for those reasons.
Someone raised a point here that once a client is denied DHCP it will give up retrying connection to same SSID.
I am trying to understand if in the case of Max Associations, client retries (to the same SSID) and eventually he connects to another AP.

I am trying to understand if in the case of Max Associations, client retries (to the same SSID) and eventually he connects to another AP.
Generally yes, I believe.
Also, if you try to limit by DHCP you have several other problems;
- IPV6
- need a separate network segment
- will break existing connections
- some devices may just keep trying and trying...
Max Associations would be the most effective solution then.
Strange this config is not available in home wifi routers.
Even DIR-825 which is otherwise feature-loaded doesn't have this.

Strange this config is not available in home wifi routers.
Home routers are designed as standalone units for home usage. That is 25 clients tops without roaming.
If you want to cover a congress hall you need to move to dedicated APs for Business.

dedicated APs for Business
Or OpenWrt

Or OpenWrt
While I understand OpenWRT will provide the feature, I see issues when running it on all APs.
In many cases we re-use the hardware already available, here not many are openwrt compatible.
In such cases I can buy just one new device for "router" and re-use remaining as AP.
Now If I am buying everything new, then I would buy APs OpenWRT compatible; however even among those there are many caveats on wireless side. I may be wrong but I felt wireless works best on stock firmware. Yes, may be Archer C7 is good with OpenWRT, but it's costly and overkill as AP.
So I am gravitating towards a single powerful OpenWRT device as router (to get the programmable goodness of linux) and remaining as APs with stock firmware.

So I am gravitating towards a single powerful OpenWRT device as router (to get the programmable goodness of linux) and remaining as APs with stock firmware
It's definitely a reasonable strategy, but you might look at something like the TP-Link eap 225 which maybe isn't quite as configurable as the ubiquiti gear, but it's cheaper and has quite a few configs.
eap 225 is a valid candidate.
However so far I have compared it with a consumer router like Archer c60 and I have questions why eap is 4 times costlier.
Both are ac1350 and both provide MU-MIMO.
Archer C60 port speed is 100 Mbps vs eap-225 Gbps.
Price ratio is 1:4.
What else is gained from eap 225 vs consumer grade routers like c60 (apart from the configurability).
Will it support more users or range?
Can one eap 225 replace 4 c60's?
Consumer all-in-one routers -- lots of competition, low margins, race to the bottom
Commercial APs -- limited competition, most from big-name firms, corporate budgets
Close to the same hardware in both from a vendor like TP-Link, if you check the specs on openwrt.org or wikidevi.com
(Well, maybe not for a C60 with 100BASE-T ports)
Edit: I had thought that the EAP225 was comparable to the US pricing dlakelan mentions in the next post and that the C60 was dirt cheap, in the US$20 range
Well, an EAP225 costs $70 here in the US
an archer c60 costs $63
so I'm not seeing that same dilemma.
the EAP225 offers you a "controller" based solution, where you can manage all the access points from a single control panel, and Power Over Ethernet for ease of install in a good location not near a plug, like a ceiling or on a wall.
It seems unfortunate that you have such a massive price difference where you are.

It seems unfortunate that you have such a massive price difference where you are.
In some parts of the world, the price of some devices aren't reasonable. Where I live, for example, Archer C7 is for around $100 new (and there is the factor of wages. That's, by the way, sometimes people cling to low-specs hardware, because devices are not as easy to get as they are in other countries.
Keep them on the same L2 segement/VLAN, so that clients do not need to change ip when roaming.
Install wpad-full and enable 802.11r, so that bss-fast-transition allows smooth roaming between them.
Cut power - especially on the 2.4ghz band with its high range - and try to set a min-rssi. On Openwrt this can be achieved by specifying minimum speeds (or taking them out as someone mentioned above).
On UBNT-gear you could try to hope for 802.11v and 802.11k, but that's a complicated story.
Check the threads that already exist on this topic:
You could use 802.11k for asking the client what is in his range and send some 802.11v fast bss transition to the client to switch to 5Ghz. Then 802.11r would work too.
There are different milestones you have to do:
Network-Manager Patch has to be accepted for supporting 802.11r (if u use network-manager...)
Fast BSS Transition be reachable via ubus
When some daemon could be implemented independent of the hostapd. Personally, I think this would be the nicest solution.
I expe…
While testing SQM, I see that there is a major difference between the 5,5 and the 2,4 Mhz Interface, not only regarding the speed. The clients connect to the one or the other, depending on whatever I dont know... I have set up the same (E)SSID and BSSID for both. For the single clients it is important to connect to 2,4 GHz only connect to 5,5 GHz if reachable connect to any Is it possible to define the relation of client -> (preferred) interface somehow on OpenWrt? Edit: Or is the propose…
Good question! Yes, this is what I suggest. I use the following scheme: SSIDfoo_2.4 SSIDfoo_5.4 If a device is capable of 5.4 GHz, I only configure it for 5.4. In most cases, 5.4 will be less congested and have higher bandwidth. This doesn't work on some OSes. It will see 2 BSSIDs (MAC address of the APs) and randomly select.
Looks like a patch is required for MIN RSSI:
You could go to shorter polling or fewer strikes, though that is likely to cause more unwarranted disconnects.
When a mobile phone or tablet is not actively exchanging data, it puts its wifi system to sleep. It may not transmit any wifi packets for a minute or more. In that case in the signal poll, the AP's wifi driver repeats the last received signal report, as it is the only information available but will be completely wrong if the phone or tablet has been moved.