Tough wireless coverage areas.

[dpwifi]

Hi All
I have been having issues closing the last gaps of our wireless coverage. Its 10 buildings with roughly 280 APs. We have several users who experience slow performance. I chalk it up mostly to the high level of noise and the general war zone that is the 2.4 GHz band. It’s a dense AP deployment and with a 3 channel plan I know there will always be a level of co channel interference and so on.

My question is has anybody had any luck with the noise immunity feature being turned up to level 3?

Also when band steering gets turned up to force 5ghz in a attempt to push more clients on the A radios many clients can’t associate to the access point they were previously on or any in the area.

[cjoseph]

It would seem that you have a number of issues. Please open a support case so that they can all be discussed in a timely manner with someone who would have access to all of your information. We can give you general information here, but there is a limit to how much personal info we can get from you, and that will hinder you finding a solution.

With that being said, you can increase the interference immunity, but that will also decrease the coverage of the access point at the same time. If you have a dense deployment and you cannot eliminate that non-802.11 source of interference, you can certainly try this. Once again, you should only change this knob based on the advice from the good people of TAC.

[dpwifi]

I do have a call open with them but they are at the limit of what can be done. We have various settings on and off.

Just wanted to put up a post here and see if anyone had suggestions about the issue with band steering.

[cjoseph]

---

Hi All
I have been having issues closing the last gaps of our wireless coverage. Its 10 buildings with roughly 280 APs. We have several users who experience slow performance. I chalk it up mostly to the high level of noise and the general war zone that is the 2.4 GHz band. It’s a dense AP deployment and with a 3 channel plan I know there will always be a level of co channel interference and so on.

My question is has anybody had any luck with the noise immunity feature being turned up to level 3?

Also when band steering gets turned up to force 5ghz in a attempt to push more clients on the A radios many clients can’t associate to the access point they were previously on or any in the area.

---

What version of code is this?

[dpwifi]

Aos-w 3.4.3.2

[cjoseph]

---

- To get a definitive idea about what is going on in your environment you would need a packet capture as well as a spectrum analysis to get a definitive idea what is going on in your environment. Without that information collected an analyzed, it is hard to say definitively what is going on in your environment.

- The band steering algorithm has improved Since October 2010 when that version of code came out, but there is no saying how much it can/cannot improve what you are currently experiencing. An upgrade in the same train of code should be considered.

- Presentations from our latest Airheads in Dallas are in the thread here: http://airheads.arubanetworks.com/vBulletin/showthread.php?t=3279 The file also contains a presentation called "TAC top 10" which, starting on slide 40 lists troubleshooting steps that you can take to narrow down your issue, as well as what information you should have ready for TAC. Use that in conjunction with TAC to figure out what your issue really is.

[dpwifi]

I have looked over the area with a spectrum analyzer and I know the buildings inside out. We turned client aware off to allow the APs to better change channels for the best layout but it’s so dense it hasn’t seem to have a effect. Users don’t seem to notice the AP’s changing channels more liberally, so we left it off.
Also your usual things are in effect like, drop broadcast and multicast, convert broadcast to ARP requests, turn down the power of the radios but the buildings are bunkers loaded with outside interference. Noise immunity is the last of the knob adjustments I could make to hope to see some improvement.

The rf neighborhoods work as they should along with load balancing, the math works out when you look at how the APs are choosing channels.
It’s become increasingly frustrating for me as it is appearing that the group of users that are experiencing poor coverage is mostly due to their wireless cards being an atheros that aren’t too powerful and more for home environments then the one we have.

[dpwifi]

Thank you I appreciate your help, as I have had help with some of your other posts in other threads.

[cjoseph]

What I would do is focus my attention on a single client that has an issue. Do packet captures on that client's channel; set up user debugging for that client. Gather statistics for that single client. Do throughput tests for that client. Gather statistics for that client's access point, as well as the switchport that the access point is connected to. Collect up all of that data and have it analyzed.

It is MUCH easier to focus on a single client's experience and that client's individual performance issues than try to affect many other clients. Ping and throughput tests in that location on that client should be compared with ping and throughput tests with that same client in a different location to narrow down location-specific issues.

Airwave is a good tool to collect RF trending information about access points, controller, as well as that client and will allow you to correlate instances of degradation with historical fingerprints of that environment. If you have Airwave, please update it to the latest version, because it collects additional RF information that older versions do not. If you do not have Airwave, please contact your Aruba account team so that you can evaluate it. It can literally collect the majority of the datapoints necessary to narrow down your issue.

[dpwifi]

I have there isn't anything unusual about it. It's a good suggestion but I have worked with one very nice client who can't get good performance out of the room. The AP data is no different then where anywhere else. Maybe a higher number of CRC Errors and I have tons of data from that one client I could poor into this thread or with my TAC call but the issue remains the same. The user works great if they connect to the A radio. Very poor if they connect to the G radio.

Iperf tests are poor, and the spectrum is full of noise and high RSSI of outside interferers

Logic would dictate that if one area is bad and on the other side of the building its good there has to be something in the controller logs or the ap data. But I look at all ap's in the building and don't see anything, no common error, etc

I do

Show ap debug client table
show ap association ap-name
show ap debug system status
show ap arm rf-summary
show ap arum neighbours
show ap debug clien stats

and so on, thats cutting it short. Nothing really stands out. It's shaping up to be a dense deployment in a tough environment with a noisy 2.4ghz spectrum.