Band steering - "priority" to 5Ghz

RouterOS Wireless Networking
1

Hi everyone,

I’ve read several other topics about band steering and 5Ghz priority here on the forum. So I quite have the picture.

Ever since I have my Chateau12 I wondered, why so many clients register with wlan1 (2.4Ghz) and hardly none at wlan2 (5Ghz). But I did not research, because there were other fields in ROS 7 that got my attention.

So now I thought it is time to find out whats happening. So I opened the wireless network scanner on my macbook to find out, why it connects to 2.4Ghz instead of the faster 5Ghz.

Please see it as an example:

RSSI of 2.4Ghz: about -50
RSSI of 5Ghz: about -70

Where -70 is just a decent signal for 5Ghz, the client uses 2.4Ghz instead. After reconnecting Macbook-wifi, it first connects 5Ghz sometimes, but switched to 2.4Ghz after some 1 or 2minutes.

I used Wifi-analyzer for Android, in most rooms/places where I use wifi-devices, I have round-about -70RSSI for 5Ghz.

Mikrotik features no band steering.

First approach:
I added a wireless access lists to prevent clients to access wlan1 (2.4Ghz) in a signal better than -70RSSI. But that kind of sucks, as some of my clients are 2.4Ghz only (like Hap lite, map lite) and would not connect anymore. And I do not like to add some “whitelist” just for these devices. So I reverted the whitelist stuff again.

Second approach:
Lower tx-power of wlan1. I increased antenna-gain to 10. Now wlan1 transmits at 10dbm (was 17dbm) before. 2.4Ghz wireless signal got a little weaker, but still it has significant better RSSI than 5Ghz. But it seems, that some clients now already switch over to 5Ghz. But just a few. On the other hand I like to have wifi-signal when I am outside the house and/or at a far away room. So I’d like to keep TX-power at it’s regulatory-maximum.

Has anyone a better idea? Anyone has a script, that emulates some kind of “poor man’s band steering”?

It really annoys me, that sitting away like 5 meters from my Chateau12, that devices use 2.4Ghz. That’s not nice and needs at least an improvement - when there is no real solution available.

Thanks.

2

If you have already read all other topic, why open anoter one?

Except what you already have read, there is not a solution (for now, until mikrotik does something)

3

Because I did not want to hi-jack other topics. Some OPs have other requirements like “my 2.4 is so crowded. I’d like to limit clients or force them to 5g”. Some are about capsman. Some topics date back to 2017. I did not want to bring up my question in a topic with already 100 comments.

4

All what Mikrotik does is ©normis: “but when 5g and 2g signal is same, then apple chooses 5g”

But that is academical. When there is a simple obstacle (NOT even a wall) in router direction - 2g has better RSSI.

5

I prefer to use your second approach and add more accesspoints. Especially outside! And both your devices but also your neighbors will prefer it.

6

Use more APs. Shut of 2.4 on most of them. Turn down the 2.4 on the working ones. Bounce the 2.4 off and on for a few seconds at some interval.

This is how I managed to steer devices onto 5Ghz if they have it.

I started making this script that would add devices to and ACL once they connected to 5GHZ. Then I was going to use a rule that would check that ACL vs the signal on 2.4 and boot it…

Then Apple/Google/Microsoft started adding the “private random Mac address” BY DEFAULT!!! That shot down my plans…

7

2.4ghz down to 10dbm. 5gzh at 21dbm. But still only ONE device connects to 5ghz. The others all go to 2.4ghz. But Mikrotik does not see any problem…

8

Given that reducing TX-power of wlan1 to 10dbm did not improve the situation. Now I am back at a simple access-list solution. Not that optimal - but it works for now. I refuse to connect to 2.4ghz if signal is better than -70. It maybe could work with -60 or -65 too - but I need to slowly go down and watch the behaviour.

I have only 2 devices that only support 2.4ghz band. HAP-lite and MAP-lite. The map is quite far away and has a signal of -75 due to the distance itself. The HAPlite is whitelisted.

/interface wireless
set [ find default-name=wlan1 ] default-authentication=no

/interface wireless access-list
add allow-signal-out-of-range=30s interface=wlan1 signal-range=-120..-70
add interface=wlan1 mac-address=08:55:31:E4:E4:C9 comment="haplite 2ghz only"
9

One thing I noticed/observed:

Some of my clients (Android, Macbook) first connect to 5ghz and switch over to 2.4ghz within approx a minute after connecting to 5ghz.
For me this looks like, that these devices prefer 5ghz by default. They connect. Then they find 2.4ghz with “better” signal and switch over.

–
Has someone a script, that can handle following scenario?

Pseudocode:

IF device already connected to wlan2 AND want to connect to wlan1 AND wlan2-signal is better than -80: DENY connect to wlan1
10

First the client often has option to select “preferred band”.

Then what you need is Access rules per Wifi interface.
Example: client A has at same location 70 RSSI in 5Ghz and 55 RSSI in 2.4
→ Allow connection to 5Ghz -120..120 (it can always try to connect to 5G)
→ Allow connection to 2.4Ghz -120..-65 (only if signal is getting below -65 (which then would be ~-90 in 5G)

This means 2.4G will only be used when signal becomes pretty bad.

Inconvenience is, you need to do this per client as signal levels are different per client…
But for an environment with known clients and a bit of tuning, this can do quit well.

11

Isn’t that quite similar to my current access-list configuration I posted above?

12

This is how it looks like without access list rules. It is quite useless to have 5ghz enabled at all, as not a single client would connect to it. they all prefer wlan1 cause of slightly better RSSI.
Screenshot_20211130-084629.png
t

13

Do you have access to a spectrum analyser that covers both the 2.4GHz and 5GHz bands?

In free space, 5GHz suffers from around 7dB more attenuation than 2.4GHz, so you’d need to have to reduce the 2.4GHz signal by 7dB minimum just to give them the same signal strength at the client device.
Now, you’ve reduced the signal by 10dB by increasing the “gain” value on the antenna, but only a spectrum analyser will tell you whether that actually means the signal strength was down by enough at the client device, ignoring the effects of walls, bodies and everything else you find in the average home/office.

If you don’t have a spectrum analyser, you can still see what happens when you set the antenna gain back to default and reduce the TX power setting by 12 or even 15dB on the 2.4GHz radio. This is more likely to give a stronger signal strength on 5GHz at least in the room the AP is in and you should get devices preferring that band.

14

This is a difficult one to solve in wifi. Because in wifi the CLIENT decides which band to use. This is different from mobile networks (3G,4G,5G …) where the network devices decide who will serve a client at what band. “Band steering” in wifi will never be perfect. Client devices use different algoritmes for making the decision.

KIcking a client off or refusing the 2.4GHz might help, if the client learns that some AP/band has a bad connection. Otherwise it is just statistics. Flapping between bands is also a common problem for some clients (Some smartphones tend to swap after each sleeping period. Sometimes you see 4 second connections.). The algoritmes are unknown and divers.

The “station roaming” is more active with lower signal level. Adding more AP and giving better coverage reduces the roaming attempts. (MT station can switch this on or off. The default behaviour has changed from on to off in RouterOS. With “off” the station will not be scanning for a better AP in the background, as can be followed in the “wireless,debug” logging.

One way to mitigate is adding a second SSID that is band specific. For stable and well behaving clients the common name can be used. But again be carefull as the client might swap to the other SSID name if it is a known network.

How can we inform the client what we want? Maybe (maybe!) abusing the Hotspot 2.0 “Interworking profiles” might be a way to inform the clients what we want, as 802.11k 802.11v is missing.

Some reading: https://www.networkcomputing.com/wireless-infrastructure/4-wifi-band-steering-myths

15

If you don’t have a spectrum analyser

… you might use just another MT AP in station mode to measure

http://forum.mikrotik.com/t/antenna-gain-missing-in-6-46-8/145162/1

16

bpwl
If only… I swear…

Moving on…
I think I recall a document I read a long time ago about “holding the 2.4 transmit beacon” to get the client to hear the 5 first.

The trick I tried for a little bit was making an ACL that kicked off MAC address groups that I “figured” had 5 GHZ radios is the 2.4 signal range was above 75. I got the idea when I was trying to make URC remotes connect to the right room when you move around.

Alas the 2.4 radios dropping or no longer passing traffic while still showing connected in caps man wasted a ton of my time.

Since I gave up on caps-man for having 2014 ACv1 drivers in radios today… I stopped trying to pick a turd by the clean end.

17

Since I gave up on caps-man for having 2014 ACv1 drivers in radios today… I stopped trying to pick a turd by the clean end.

Yes it is not easy. The weapons we get are very limited. One can play with minimal RSSI, and with higher basic rates. Just kicking off clients from the wifi does not instruct the client on what else to do, only to try again. Higher basic rates to avoid clients with disturbed reception to keep trying to reconnect. (Just kicked out my friends airco unit from his AP with MAC address and access list, The espressive system in there now attempts reconnection 3 times per second.)

Is RSSI the critrerion for the access list we want? Maybe not, I have BYOD clients connecting at 54Mbps @ -86dBm, and at 12 Mbps @ -72 dBm but poor CCQ. Saving AP beacon broadcast time? Kicking off sticky clients? Well maybe yes. What I want is preserve air-time, and use it with the best possible throughput. So that 12Mbps was even worse with its 3 fold average retransmit.

2014 ACv1 drivers? Optimistic? I rather see “802.11n” drivers adapted for “ac” hardware, enough to make it work. A-MSDU and A-MPDU buffer sizes are too small for good AC performance, but the needed RAM may be missing. And newer “ac” developped aditional adaptive weapons are missing. So we have to fight with what we have, and use Dog Poo Bags to keep the hands clean.

18

And you still refuse to say GARBAGE!

As to your comment about “going to battle”…

Mikrotik gave me cotton candy, to throw at steel tanks.

I swear… You are clearly better than I am at wireless… If you would just publiclly call Mikrotik wireless for how deeply flawed it is… And stop apologizing for it… Maybe Mikrotik might actually address things???

I had to go back and forth with them for months (and had to call them out here when weeks would pass between emails) before they copped to the fact that it was their wireless causing all my customers problems.

It cost my company thousands to replace all those radios.

19

It’s all point of view.

Majority of people using these devices do not have access to internet allowing to stress those devices enough.
There are plenty of cases where this underperforming WiFi is more then sufficient.

E.g I have two cap devices I use in a rental home. I don’t care the speed never goes beyond 100mbps. Main line is at best 15 mpbs using SXT LTE !
It’s located in the middle of nowhere. Half of the time literally no cell reception, not even for phone calls.
It does not matter to use more expensive and maybe less easy to handle equipment from another vendor.

Not saying the performance problems do not need to be fixed, they certainly do, just giving some perspective. For the majority of customers, there is no problem. They just need a working wifi. Period.

20

I have to agree with holvoetn.

I have 300Mb down service and I am using the cAP ac in my home, managed by capsman. None of my wireless use-cases require massive throughput, I use wireless on my family’s phones and tablets, where the most stressing traffic is some video streaming and some Zoom usage. None of that needs huge WiFi throughput.
For laptops, the same thing applies except when I need to work with larger video files. For this I use a wired GbE connection on my dock anyway.
Even downloading large files like OS or software updates isn’t a problem over a sub-150Mb WiFi connection as I’m never in a position where if something doesn’t download in 37 seconds flat the earth will stop turning.

Yes, Mikrotik should address the issues , but even then I believe they should focus on stability in crowded spectrum over pure speed. Other than for speed tests and pure bragging rights, on WiFi speed is rarely the most important part of a connection, stability and being able to support more clients is much more important for most people.