Roaming between AP's fails creating lengthy reconnections (including full DHCP process)

Roaming between AP's fails creating lengthy reconnections (including full DHCP process)

Roaming between AP's fails creating lengthy reconnections (including full DHCP process)
Roaming between AP's fails creating lengthy reconnections (including full DHCP process)
2024-06-11 02:57:50 - last edited 2024-07-15 14:39:44
Tags: #Roaming
Model: OC200  
Hardware Version: V2
Firmware Version: 5.13.30.20

Symptom: Cellular WiFi calling drops calls.  WiFi phones using conference or phone software (like WhatsApp) show reconnecting for as much as 10 seconds when moving between AP's.  

 

System consists of 10 access points inside and around resort buildings operating through 3 Omada managed switches, ER7206 GW and OC200 controller. 9 of the 10 AP's are hardwired, only one is mesh connected.  AP signal overlap is plenty to facilitate seamless roaming.  802.11r is enabled, as are Advanced Features and all the roaming checkboxes.   

 

A ThinkPad laptop (.50), and Samsung SmartPhone (.46) were used to test roaming.  Laptop Wireshark capture shows DHCP process invoked for a reconnection.  Smartphone experiences failed authentication and multiple DHCP attempts.  Omada log confirms and shows multiple instances of DHCP issuance. In the worst cases, phone test devices running WiFi Man shows dragging AP's too far when stronger AP's are available. It is as if a neighbour list is not advertised properly.    

 

Omada Logs show many WiFi mobile devices randomly failing authentication during roaming.  The WiFi client device knows the SSID and PW... how does this fail authentication?  Isn't "fast roaming" supposed to reduce the number of authentication steps when moving between AP's?  There is one EAP110 outdoor that is worst for this "fail authentication" issue, but it occurs everywhere in the site.  

 

Roaming just seems broke!   Log Sample: 

 

Galaxy-S20-Ultra-5G is roaming from Dom2 Lopa[Channel 153] to Marko Garage[channel 6] with SSID Dom3  Jun 10, 2024 21:40:01    
ThinkPad-T470p is roaming from Dom3 West Wall[Channel 1] to Marko Garage[channel 6] with SSID Dom3  Jun 10, 2024 21:40:01    
DHCP Server allocated IP address 192.168.33.50 for the client[MAC: 2e-bf-f0-36-09-f7].  Jun 10, 2024 21:40:00    
DHCP Server allocated IP address 192.168.33.46 for the client[MAC: e4-70-b8-53-6f-d3].  Jun 10, 2024 21:39:58    
DHCP Server allocated IP address 192.168.33.46 for the client[MAC: e4-70-b8-53-6f-d3].  Jun 10, 2024 21:39:55    
DHCP Server allocated IP address 192.168.33.46 for the client[MAC: e4-70-b8-53-6f-d3].  Jun 10, 2024 21:39:53    
ThinkPad-T470p is roaming from Dom1 West Wall[Channel 1] to Dom3 West Wall[channel 1] with SSID Dom3  Jun 10, 2024 21:39:31    
[Failed]Galaxy-S20-Ultra-5G failed to connected to Dom2 Lopa with SSID "Dom3" on channel 153 because WPA Authentication failed.(1 time in a minute) Jun 10, 2024 21:39:30   
 

Cellular WiFi calling doesn't tolerate connection gaps and requires seamless handoffs between AP's.  Some IP TV's will buffer and hang, which I suspect happens when on the edge between AP's and roaming happens... that seems to invoke a full reconnection, instead of "fast roaming".  There are other "Poltergeist" issues that I've been monitoring to determine if it is the user equipment, or a TP-Link issue. 

 

Any insight appreciated.    

 

Tony 

  0      
  0      
#1
Options
1 Accepted Solution
Re:Roaming between AP's fails creating lengthy reconnections (including full DHCP process)-Solution
2024-07-15 14:39:40 - last edited 2024-07-15 14:40:40

After a couple weeks monitoring the situation, roaming works a lot better with 802.11r DISABLED. Since the site is not 802.11x, seems that turning on 802.11r causes problems as noted above in this thread and is detrimental for basic authentication. I'm not seeing the DHCP flurry like before. Roaming standards k and v do their job nicely. 5 GHz is listed as preferred. I'm seeing most SmartPhones are attempting to stay connected at 5 GHz. Only where the user equipment (UE) link is stretched do they roam to 2.4 GHz... but come back up to 5 reasonably soon. WhatsApp and other social media phone apps work fine. Cellular WiFi calling is doing better now, but can still drop as a result of WiFi roaming. Seems cellular WiFi calling is very sensitive to upset. Thank you @Fae for your solution vector.

Recommended Solution
  0  
  0  
#7
Options
6 Reply
Re:Roaming between AP's fails creating lengthy reconnections (including full DHCP process)
2024-06-12 03:43:08 - last edited 2024-06-12 03:44:05

Hi  @RF_Dude 

 

May I have the SSID settings, and EAP model&firmware version you are testing?

 

Please try disable PMF on your SSID advanced settings, the WPA mode choose WPA2-PSK-AES, don't use WPA3. 

 

For Apple devices only, please also disable 802.11r.

 

Please disable 2.4Ghz for testing, since 2.4Ghz can be easily disturbed.

 

For your reference, "WPA authentication timeout" will happen not only when the device has wrong password, it also happens when "the device does not have a good wifi signal".

RF_Dude wrote

Symptom: Cellular WiFi calling drops calls.  WiFi phones using conference or phone software (like WhatsApp) show reconnecting for as much as 10 seconds when moving between AP's.  

 

System consists of 10 access points inside and around resort buildings operating through 3 Omada managed switches, ER7206 GW and OC200 controller. 9 of the 10 AP's are hardwired, only one is mesh connected.  AP signal overlap is plenty to facilitate seamless roaming.  802.11r is enabled, as are Advanced Features and all the roaming checkboxes.   

 

A ThinkPad laptop (.50), and Samsung SmartPhone (.46) were used to test roaming.  Laptop Wireshark capture shows DHCP process invoked for a reconnection.  Smartphone experiences failed authentication and multiple DHCP attempts.  Omada log confirms and shows multiple instances of DHCP issuance. In the worst cases, phone test devices running WiFi Man shows dragging AP's too far when stronger AP's are available. It is as if a neighbour list is not advertised properly.    

 

Omada Logs show many WiFi mobile devices randomly failing authentication during roaming.  The WiFi client device knows the SSID and PW... how does this fail authentication?  Isn't "fast roaming" supposed to reduce the number of authentication steps when moving between AP's?  There is one EAP110 outdoor that is worst for this "fail authentication" issue, but it occurs everywhere in the site.  

 

Roaming just seems broke!   Log Sample: 

 

Galaxy-S20-Ultra-5G is roaming from Dom2 Lopa[Channel 153] to Marko Garage[channel 6] with SSID Dom3  Jun 10, 2024 21:40:01    
ThinkPad-T470p is roaming from Dom3 West Wall[Channel 1] to Marko Garage[channel 6] with SSID Dom3  Jun 10, 2024 21:40:01    
DHCP Server allocated IP address 192.168.33.50 for the client[MAC: 2e-bf-f0-36-09-f7].  Jun 10, 2024 21:40:00    
DHCP Server allocated IP address 192.168.33.46 for the client[MAC: e4-70-b8-53-6f-d3].  Jun 10, 2024 21:39:58    
DHCP Server allocated IP address 192.168.33.46 for the client[MAC: e4-70-b8-53-6f-d3].  Jun 10, 2024 21:39:55    
DHCP Server allocated IP address 192.168.33.46 for the client[MAC: e4-70-b8-53-6f-d3].  Jun 10, 2024 21:39:53    
ThinkPad-T470p is roaming from Dom1 West Wall[Channel 1] to Dom3 West Wall[channel 1] with SSID Dom3  Jun 10, 2024 21:39:31    
[Failed]Galaxy-S20-Ultra-5G failed to connected to Dom2 Lopa with SSID "Dom3" on channel 153 because WPA Authentication failed.(1 time in a minute) Jun 10, 2024 21:39:30   
 

Cellular WiFi calling doesn't tolerate connection gaps and requires seamless handoffs between AP's.  Some IP TV's will buffer and hang, which I suspect happens when on the edge between AP's and roaming happens... that seems to invoke a full reconnection, instead of "fast roaming".  There are other "Poltergeist" issues that I've been monitoring to determine if it is the user equipment, or a TP-Link issue. 

 

Any insight appreciated.    

 

Tony 

 

 

>> Omada EAP Firmware Trial Available Here << *Try filtering posts on each forum by Label of [Early Access]*
  0  
  0  
#2
Options
Re:Roaming between AP's fails creating lengthy reconnections (including full DHCP process)
2024-06-12 15:17:15 - last edited 2024-06-12 15:24:00

Outdoors it is mainly EAP225-Outdoor. (5.1.6). We have one EAP110 (5.0.7) at one extreme end of the system. Indoors, it is a mix of EAP245 (5.0.3... but one new 245 is 5.0.5) in two buildings, and EAP650 (1.0.13) and 655 (1.2.6) in one building. All are the latest firmware as made available through Omada Check for Update. Four SSID: DOM2 is native management LAN. DOM2 MOOCH is a guest network, And DOM3 delivered as VLAN 33 as is DOM3 MOOCH for guests. I have 802.11r turned ON.

 

Per your suggestion, will turn 11r OFF. There is no info how to use PMF, so this was checkbox CAPABLE. I will DISABLE. Under Advanced Features, all the checkboxes are enabled: Fast Roaming, AI Roaming, Dual Band 11k Report, Force Disassociation and Band Steering set to Prefer 5 GHz. WPA-Personal with Mode set to WPA/WPA2-PSK/Auto MLO is disabled.   

 

It is likely and expected that in some handoff cases, devices will downgrade to 2.4 GHz before roaming to a new AP. Roaming between 5 GHz AP's might be successful inside buildings due to overlap, but is not universal everywhere due to obstructions, walls, etc.   

  0  
  0  
#3
Options
Re:Roaming between AP's fails creating lengthy reconnections (including full DHCP process)
2024-06-21 13:54:51
From what I'm understanding, for a typical residential (not business... not 802.11x) scenario, only k and v are used for fast roaming.  And of these, "v" provides or directs a best candidate suggestion to a User Equipment (UE).  To troubleshoot roaming, I believe I'd have to confirm that "v" is working, it is suggesting the best AP candidate, and the UE is acting on the new AP suggestion.  For instance, the UE might be acting on it, but is too weak to associate with the recommended AP, resulting in "Failure to Authenticate" logs in Omada. While I'm using WiFiMan to display UE activity and roaming, I could not find a tool to use on a phone to display 802.11 management packets like v and k. Anyone have any suggestions? Wireshark can be made to offer this on a laptop, but it seems many wireless adapters do not allow management monitor. Yeah... get an adapter that does. Obviously, troubleshooting this to a conclusion is itself a learning opportunity. Thanks for vectors pointing towards solutions.
  0  
  0  
#4
Options
Re:Roaming between AP's fails creating lengthy reconnections (including full DHCP process)
2024-06-24 01:29:30

Hi  @RF_Dude 

 

I don't have any better suggestions on Android/Phones APP(that shows roaming process). 

 

I had an idea about why some of your devices may downgrade to 2.4Ghz before roaming. I believe it is caused by the high transmitting power of outdoor AP. Outdoor version has higher transmitting power, if the "neighbor AP" is too far away, the device may switch to 2.4Ghz so it can connect the AP in a farther distance. 

 

You can try to change the transmit power of 2.4Ghz to "Medium" or "Low", to improve the roaming experience. 

 

>> Omada EAP Firmware Trial Available Here << *Try filtering posts on each forum by Label of [Early Access]*
  0  
  0  
#5
Options
Re:Roaming between AP's fails creating lengthy reconnections (including full DHCP process)
2024-06-24 17:32:02
@Fae, you are right, to a point. I would think that we can design and provide for sufficient 5 GHz signals indoors, but there will always be transit areas that people find where they roam to 2.4 GHz. Outside, SmartPhones stay on 5GHz if LoS, but pass some tree's or slightest obstruction, and the phone ends up on 2.4 GHz. Outdoors, we do have 2.4 GHz overlap, but not continuous 5 GHz coverage. Very sharp, like a knife where 5GHz works, and doesn't. The AI AP Optimization largely gravitates towards full power, even indoors where AP's are down the hall from each other. I have manually adjusted down 2.4 GHz power, but have to leave power in CUSTOM mode, else AI algorithm puts it back into full power.
  0  
  0  
#6
Options
Re:Roaming between AP's fails creating lengthy reconnections (including full DHCP process)-Solution
2024-07-15 14:39:40 - last edited 2024-07-15 14:40:40

After a couple weeks monitoring the situation, roaming works a lot better with 802.11r DISABLED. Since the site is not 802.11x, seems that turning on 802.11r causes problems as noted above in this thread and is detrimental for basic authentication. I'm not seeing the DHCP flurry like before. Roaming standards k and v do their job nicely. 5 GHz is listed as preferred. I'm seeing most SmartPhones are attempting to stay connected at 5 GHz. Only where the user equipment (UE) link is stretched do they roam to 2.4 GHz... but come back up to 5 reasonably soon. WhatsApp and other social media phone apps work fine. Cellular WiFi calling is doing better now, but can still drop as a result of WiFi roaming. Seems cellular WiFi calling is very sensitive to upset. Thank you @Fae for your solution vector.

Recommended Solution
  0  
  0  
#7
Options

Information

Helpful: 0

Views: 484

Replies: 6

Tags

Roaming
Related Articles