VoIP over WLAN - Client Roaming Settings

[Frank Wagner]

Hello,

what would you suggest, are the best settings in the "Client Roaming Section" of the WLC Configuration for the following requirements:

VoIP over WLAN deployment, 802.11a, WLC 4402, 25 - 30 access points (mainly 1252, a few 1130), 7925G phones.

In the moment the settings are set to default, that means: Min. Rssi = -85; Hysteresis = 3; Scan Tresh. = -72; Trans. Time = 5

Unfortunately we are not completely satisfied with the roaming behavior of our phones, so there is the question,

if the roaming behavior of the phones can be improved. Sometimes it looks like the phone takes too much time to make a roaming decision.

Thanks,

Greeting Frank

[mlieber · Accepted solution]

Hi Frank,

it seems that we have exactly the same issue.

WLC5508, SW 7.0.98

AP1140 and 1250

Phone 7921 and 7925 and 1.4.1 SW Version

Did you have got any news from Cisco about your TAC case?

Best regards

Martin

[Frank Wagner]

Hello Martin,

Unfortunately our TAC case was closed, because at one point last year,

we were not been able and willing to produce even more and more logs and debugs

an invest more and more time without getting any result.

At the moment we are considering to replace the whole VoWlan infrastructure,

or to try one last step in co-operation with an big Cisco distributor here in Germany.

Greetings Frank

[mlieber]

Hello Frank,

it seems that we have resolved the roaming issue.

QoS configuration is very important especially between the switches. Additional the QoS mapping is also very important.

About your question in the beginning: Configure CCX Client Roaming Parameters like minimum RSSI, Hysteresis etc.

We have changed these parameters and we are sure that this changing has improved the quality and reduced the roaming delay very strong.

Another important part are the phones.

We have in the environment 7921 and 7925 phones.

We can take a call with the 7921 phones in areas where it is not possible with the 7925 phones.

It seems that the quality of the 7921 phones are much better than the 7925 although the RSSI value in the phones site survey menu is almost the same.

What I have to say is: The RSSI value in the site survey menu of both phones is -65dbm. I cannot take a call with the 7925 phone but I can take  a call with the 7921 phone.

greetings Martin

[jhoffman]

Here's where we are at our Hospital rollout of new building with 300 new 7925 phones ( we presentaly have 7921 phones deployed).We have a  very dense "A" grade voice deployment( 30 foot on center separation between APs, up to 10 floors) We have the same issues listed in this post, a lot of time by our staff, Cisco consultant (on our dime) Cisco TAC...DeBugs, Sniffer traces, WLC Analyzer scrubbed (5508 & WISM) you name it we done it.... So today got to another TAC 2nd level engineer ran thru the usual troubleshooting steps with one more difference lets try this. Turn off "A" 36 - 54 data rates(upper 3 data rates). (global setting so it will affect everything on your "A" network  WOW\Phones, etc). This worked for 7925 intra-controller roaming. We have had our AP's approx 150 in new building on one 5508 as a work around (this works but I have x 700 plus AP in my main campus buildings so the one controller solution we found wasn't the answer. So if you have a TAC case going ask them to ref case SR 616692741 - Roaming Issues with 7925 phones. We view this a workaround as we have 300 plus WOWs on our "A" network and the global change will step on their thruput.

[Jacob Fussell]

From a number of vowlan roaming cases I've learned that keeping an 8-12channel plan in 5Ghz and having no less than power level 2 on the APs shows the best results for roaming effectiveness. The roaming/scanning logic in the 792x phones was completely re-written in the 1.3.4 software to accomodate interband roaming. We have seen where this new design can in some instances take longer to get around to scanning some channels before roaming.....it would be interesting to have 1 792x in your same RF domain using 1.3.3 as a test.

Sent from Cisco Technical Support iPhone App

[jhoffman]

Thanks for the email... been working on this problem for a while with no luck... A lot more testing on Monday to verify intra-controller roaming. Not seeing things from\or having access to that info from Cisco TAC side I'll say that every wireless deployment is different, our extensive testing in our environment wouldn't say band plan or power was the issue. The 7921 worked, the 7925 didn't. We found that if we had one floor deployed approx. 15 Aps things worked fine or if we put all APs (floors) on one controller it worked, as soon as we added more floors (APs) to more than one controller the intra controller roaming stopped working. So we view it as a density issue. (too may hot Aps to choose from) ....

To answer your post:

7925 were at the latest release 1.4.1 code before this work around... Power is automatically set by the Cisco controllers (don't want to manage\manually set 700 plus Aps) we aren't seeing that high of power level (2) decisions made by the controllers and with our Pico cell deployment not sure how power levels that high would work...

From what I know about the 7921 vs 7925 the 7921 has 2 antennas (diversity) 7925 has one.

We like to see Cisco get involved in a much greater way as this is just a work around not a solution turning off "A" 36 - 54 data rates.

Joe

[Aaron]

> The 7921 worked, the 7925 didn't.

The two phones are almost identical.  The only significant differences are Bluetooth in the latter (possible interference source) and antenna diversity in the former (possible challenge for the 7925 in high multipath environment.

> We found that if we had one floor  deployed approx. 15 Aps things worked
> fine or if we put all APs (floors)  on one controller it worked, as soon as we

> added more floors (APs) to  more than one controller the intra controller

> roaming stopped working.  So we view it as a density issue. (too may hot Aps to choose from) ....

Hm.

> To answer your post:

> 7925  were at the latest release 1.4.1 code before this work around... Power  is

> automatically set by the Cisco controllers (don't want to  manage\manually

> set 700 plus Aps) we aren't seeing that high of power  level (2)  decisions made

> by the controllers and  with our Pico cell  deployment not sure how power

> levels that high would work...

Pico cells are not what you want with 792x phones (unless the phones are stationary).  With 5GHz power levels below say 11dBm,and with mobile phones, the phones can walk right out of cell and across another cell before the phone can scan for the new APs.  Result will be intermittent audio outages of up to 5-6 seconds.

Please be aware that RRM's default power settings are optimized for 2.4 GHz (3 non-overlapping channels) not 5 GHz (8 or more nonoverlapping channels), and for stationary laptop clients not highly mobile clients.  With 5 GHz mobile clients, you want much higher power levels than what RRM gives you by default.  Some ways to manage this:

• tx-power-control-thresh set much higher than the default of -70 dBm - maybe -55 dBm
• if running recent WLC code, set a minimum power level of 2
• manual power settings (1 or 2)

> From what I know about the 7921 vs 7925  the 7921 has 2 antennas (diversity) 7925 has one.

> We  like to see Cisco get involved in a much greater way as this is just a  work around

> not a solution turning off "A" 36 - >54 data rates.

Agreed.  If using the 792x phones in 802.11a, please to be sure to use 8 - 12 channels (no more) and to have your AP power levels set to 1 and 2 only.  If there is evidence of co-channel interference (use WLC Config Analyzer to validate), then disable radios (set them into monitor mode etc.), do not reduce power.

Aaron

[Frank Wagner]

Hello Aaron,

thanks for this detailed answer and explanation,

but I've one question concerning your conclusions.

Isn't it counterproductive setting the AP powerlevel to 1 (17 dbm here in Europe)?

As far as I know the 792x phone have a maximum transmit power (in -E domain) of 16 dbm.

Greetings Frank

[wireless wlc]

Hi Guys,

  I am having exactly the same problem with wlc 5508 running 7.0.98 and 7925G fone with 1.4.1 firmware. I am not using any authentication except mac filtering on the voice ssid ! . The user is not roaming and has around 50db signal strength where he is sitting and still facing voice drops randomly. There is only LAP1242AG(with only G radio) and 7925 is set for auto-rssi (tried auto-b/g also) . Will it help if i can upgrade to 70.098.205  ? Please share any updates that you might have

regards

Joe

[Aaron]

Joe, what you are describing here is not a known issue.  What is the other end point of the call?  What is the data path between the 7925G and that endpoint?

I would recommend simplifying your test case to the minimum that demonstrates the issue of concern.  Then get simultaneous NTP synced sniffer captures from the 7925G's channel, the data path between AP and WLC, and between WLC and the other endpoint.  See where the packets are being lost.

Aaron

[wireless wlc]

Thanks for the pointers Aaron. Problem is, the issue occurs intermittently like sometimes in the morning, sometimes evening etc... so we cannot really do the captures and all when the issue occurs. When we go to site to check making many calls, there is no problem. Only mac filtering is there on the voice ssid, i did a wlc config alalyzer for the wlc configs which said its not recommended, will tat be an issue ?

Not sure if it is related, but when i manually set the channels and power, i saw the following in the logs :

*apfReceiveTask: Feb 10 11:31:53.831: %RRM-3-RRM_LOGMSG: rrmChanUtils.c:290 RRM LOG: Airewave Director: Could not find valid channel lists for 802.11bg

*apfReceiveTask: Feb 10 11:31:33.356: %RRM-3-RRM_LOGMSG: rrmChanUtils.c:290 RRM LOG: Airewave Director: Could not find valid channel lists for 802.11bg

(there are some non-cisco AP's operating on channel 11 and wlc RRM was still putting nearby AP's on channel 11, so i had to manually set those AP's to 1 and 6 channels )

regards

Joe

[Aaron]

Well ... if you can't be there when the problem happens ... then the best you can do, I suppose, is to have some instrumentation set up, so that you can look at what was going on in a post mortem.

When a problem report comes in - you can go to WCS and look at that client's history.  Look at where it was, where it had been.  Maybe you will find a pattern wrt locations or travel paths that are associated with problems.

Go back and look at alarms - interference, load, whatever - at that time.

Re MAC authentication - I can't think offhand off any reason to recommend against this configuration for voice (other than its lack of serious security value, of course.)

Re the RRM error message ... don't know of any relationship with static channel settings.  We have a bug on this:

CSCth71176 RRM LOG: Airewave Director: Could not find valid channel lists for 11b

which development closed (i.e. chose to stop working on it), because the customer who reported this stopped responding to us.  If you'd like to get this fixed, then you can open a TAC case and we can reopen the bug.  Our hypothesis was that this was being  triggered by an AP's CAPWAP tunnel flapping.  Of course, if your CAPWAP tunnels are flapping ... then that's going to cause problems for your phones.  So you'd better check that out.

Cheers,

Aaron

[Aaron]

> Isn't it counterproductive setting the AP powerlevel to 1 (17 dbm here in Europe)?

> As far as I know the 792x phone have a maximum transmit power (in -E domain) of 16 dbm.

That's a good question.  Generally speaking, from the standpoint of smooth roaming, you to use the largest possible cells - i.e. you want your access points to transmit at the highest possible power, subject to the constraint that the AP's transmit radius is no greater than the phones' transmit radius (in order to avoid one-way audio issues.)

Against the above considerations, you need to weigh traffic capacity - obviously, if you are using 2.4GHz with only 3 nonoverlapping channels, the larger the cells, the lower the aggregate call capacity in your coverage area.  So here I will assume that you are using only 5GHz with 8 to 12 nonoverlapping channels, and that call capacity is not a concern.

As you note, the 792x's maximum transmit power is 16dBm.  If you are using for example -E domain 1131AG APs, the indeed your APs' maximum power level is 17dBm.  So, with the APs at power level 1, they would be transmitting 1dB hotter than the phones ... however, the next step down, power level 2, puts the APs at 14dBm, which would give you an appreciably smaller cell radius.

In the scenario where the AP is transmitting at 17dBm and the phone at 16dBm, I think the probability of the 1dB difference inducing one-way audio is quite low.  If this is a concern, then you could set the basic rate relatively high (say 24Mbps, if your density permits), so that the phone will detect beacon loss from a relatively distant AP quickly, but allow lower rates, 12 or even 6Mbps, so that if a phone is associated to a distant AP for a bit, the packets still have a good opportunity to get through.

So - either 17dBm (power 1) or 14dBm (power 2) would be your best options - I would certainly go no lower than power 2.  If you do get co-channel interference at power 2, then turning off radios, not reducing power, would be your best next step.

Cheers,

Aaron

[peter_bruckner]

Hello,

just a short questions as far as the power options are concerned regarding active and passive gain. Does the the transmitting power of the AP (for example 17 dbm) includes the EIRP of the particular antenna AP as well? In addtione so far I was not able to find any hints concerning the EIRP of the antenna used in the7925G (EIRP according to the FCC definition: equivalent isotropically radiated power, the product of the power supplied to the antenna and the antenna gain in a  given dirction relative to an isotropic power). So question is: are there any greater differences between an AP with power 1( 17dbm) and the 7925G using 802.11a OFDM and 7925G 40 mW (16 dBm) concerning the EIRP?

Any hints&tips are welcome

Peter

[Aaron]

Peter, the configured AP transmit power does not include the antenna gain.  So if e.g. the antenna has 3dBi of gain, and if the AP radio is transmitting at 17dBm, then the EIRP (in the direction of max gain) is 20dBm.

The 792x antenna can be considered as being isotropic (i.e. having gain of 0dBi).

So, to your question:

> are there any greater differences between an AP with power 1( 17dbm) and the 7925G

> using 802.11a OFDM and 7925G 40 mW (16 dBm) concerning the EIRP?

Then if the AP is transmitting at 17dBm and has an antenna of gain 3dBi, you get 20dBm of EIRP in the antenna's max gain direction.  And the 7925G transmitting at 16dBm has 16dBm of EIRP in all directions.

However, if you look at the AP<->7925G communications, remember that the AP's antenna gain is applied in its receive path as well as its transmit path.  So there is really only 1dB greater net power in the AP->phone direction than in the phone->AP direction.  However, as the receive sensitivity of the AP is better than that of the phone, there is no problem with the AP receiving the packets from the phone.

(Bear in mind that all of the calculations above are theoretical ... in the real world, of course, your results can and will always vary.)

Aaron