Two 802.11AC access points on DFS channels

Here in the USA, I noticed if I put my two Linksys wrt3200 on a DFS channel, 100 - 144 (5GHz) then one of them will take down its connection to devices after awhile. Initially, I thought something was wrong with my router but nothing is wrong.

I'm able to set one router on a DFS channel and the other on a non-DFS channel, e.g. 36-48 or 149-161 (5GHz) or even 52-64 (5GHz).

Is this normal operation? I pretty sure in my densely populated area there are no radar emitter devices. I read up on DFS, I don't see why the latest FW 23.02 rc5 would not operate with two radio on a DFS channel.

Can you provide some clarification...

  • Are both devices set to the same channel in your initial experiment (with them both on DFS channels)?
  • In your second paragraph, you mention that you are setting one to a DFS channel and one to a non-DFS channel... is the DFS channel the same one as tested before?
  • What happens if you switch the AP/Channel association (so if router A is initially DFS and router B is initially non-DFS, make router B DFS and vice versa)?

mwlwifi is known to be problematic with DFS, which is not going to be improved - if it hurts, don't do it.

1 Like

OK, thanks for the info. I ask because I almost put my Linksys wrt3200 up for sale thinking the HW is bad. I found post on here where the problems were described; looks like the issues were just keeping the radio to a DFS channel alive.

Well in my case, a DFS channel is quite stable since I'm able to set and forget the access point is on a DFS channel. I currently operate two 5GHz access points, one on a DFS channel and the other on a non-DFS channel and never have connection problem. Actually the non-DFS channel is licensed outside the country but my wrt3200 work OK.

Scan of the other channel show neighbors with multiple radio co-channel on normal 5GHz 802.11AC.

Both AP are never co-channel (same channel). I did perform experiments that you described. The APs are set to 80MHz BW which leaves 3 DFS non overlapping channels. Anytime I put two APs on the DFS non overlapping DFS ch, one of them disables it’s radio. Swap channels same result. One AP wrt3200 in the center of my home ceiling mounted the other wrt1900 is in a detached garage 30 ft away, ceiling mounted. The AP in the house always disconnects.

When I move the in home wrt3200 to a non-DFS channel that network connection never goes away. Swap the channels on non-DFS and DFS same results, network connectivity is OK.

This tells me the radio code is seeings another AP as an “radar” interference and it must disable its radio. I don’t know, it’s all my assumption.

That's possible, but I think unlikely. IIRC, the radar signature is considerably different.

To me, this suggests that the AP in your home is likely getting a hit whereas the one in the garage is not... this could be down to the physical position of the device and the environmental conditions around them when considering the path of the radar signal. For example, 5GHz attenuates rapidly through walls... so if the radar signal would need to pass through more walls or RF dense materials to hit the AP in your detached garage relative to the one in your home, it may be that the signal has been atteunuated sufficiently as to not trigger a DFS related shutdown. If you really want to experiment, swap the two APs in terms of their locations and see if the behavior follows the locatino or the AP.

I wouldn't call it that unlikely, the law only mandates that the DFS algorithm mustn't fail to detect radar signals (no false negatives), how good the algorithm is at avoiding false positives is entirely up to the vendor (driver/ firmware/ hardware). There are quite a few specimens that are not good at this (e.g. mwlwifi or Broadcom wl/ bcm43684). I wouldn't be surprised at all, if the wrt3200acm would manifest this particular behaviour. One thing that might be interesting to test, would be configuring the same 5 GHz channel on both devices, as this might 'help' the firmware to recognize the other AP as WLAN traffic, rather than misdetecting radar impulses in the random interference/ noise of the other AP (there is always some overspill and interferences cause by other APs, even if they aren't sharing the same channels).

1 Like

good point. This would be an interesting result.

What happens if you switch the AP/Channel association (so if router A is initially DFS and router B is initially non-DFS, make router B DFS and vice versa)?

Slh and Psherman, I think your guys are correct, it looks like the AP located inside my home is seeing a detecting false radar. As you know, radar is usually a pulse (on/off switching) vs CW (continuous waveform). Weird, I consider this a "bug", addition code is needed to reboot the radio coverage is down with crontab or something similar.

I swapped router positions, move the Linksys 1900ACS (from detached garage) to center of my home and moved the Linksys 3200wrt (from home) to the detached garage. Hours later, I noticed the Linksys 1900ACS DFS 5GHz channel has no clients attached. Meanwhile the wrt3200 in the detached garage is up and running with clients attached.

As a civilian, it's probably hard to know if the radar detection is a false positive or a true positive, but this does show that the devices are operating as they should from a regulatory perspective and that the difference in location is indeed responsible for the behavior you were seeing. Although not always ideal from a utlization standpoint (depending on your neighbors/environment), the solution is really to use non-DFS channels. That should resolve the problem and keep your APs up and running without periodic interruptions.

If your problem is solved, please consider marking this topic as [Solved]. See How to mark a topic as [Solved] for a short how-to.
Thanks! :slight_smile:

This topic was automatically closed 10 days after the last reply. New replies are no longer allowed.