Wireless Access

last person joined: 6 hours ago 

Access network design for branch, remote, outdoor and campus locations with Aruba access points, and mobility controllers.
Expand all | Collapse all

Improve this terrible network design

  • 1.  Improve this terrible network design

    Posted Nov 08, 2013 03:02 PM

    A few years ago, someone at my company with little to no experience with Aruba designed the following network for a chain of hotels we service.


    • Aruba3600 in our datacenter acting as a standalone master running
    • Controller terminates around ~400 RAPS
    • Each hotel having anywhere from 10-20 AP93s all configured as RAPs back to this same controller.
    • Forwarding mode bridge since we have a dhcp server onsite which acts as our gateway and delivers captive portal as well.
    • RAP operating mode is always (regardless of connectivity to the controller).
    • Default ARM profile maxing out TX rates.
    • All users receive the default ap-role on the controller
    • 1 AP group was created as the master and all the hotels have their own ap-group which was cloned from the original so changing a setting on 1 group changes them all.
    • The controller configuration is as basic as it gets, no special tweaks nothing .
    • Every 2 rooms there is an 8" cinder block wall
    • APs are inside the suspended ceiling

    Hotels have 2 floors (some L-Shaped and some I-shaped) -- APs only on 1st floor.

    Very low or non-existent RSSI levels on the 2nd floor

    Poor RSSI levels on the 1st floors

    APs are layed out like the following image (w/ RSSI levels)



    How do you improve this without adding a local controller and not moving APs to the 2nd floor?


  • 2.  RE: Improve this terrible network design

    Posted Nov 08, 2013 03:52 PM

    I would attempt to put an AP93 in a room and determine how much horizontal and vertical coverage you would get from it.  Having access points in the halls is convenient in terms of maintenance, but has the drawback of creating RF interference, and not reaching into the rooms that they were designed to provide coverage to.


    In general, measuring and possibly moving AP93s to the rooms will possibly provide coverage to the rooms where it is needed, and avoid crippling RF interference.  The VRD for Residence Halls here:  http://www.arubanetworks.com/wp-content/uploads/NextGenAppNote_2012-06_28.pdf explains the rationale.


    The information in your post is detailed, but much more detail is required to determine if moving them into the rooms is indeed feasible.  The new design will probably require more access points, but you probably only have half as many as you will need to provide coverage to that location in the first place.  You could start adding access points to the rooms on the second floor, so that current coverage on the first floor is maintained....



  • 3.  RE: Improve this terrible network design

    Posted Nov 08, 2013 04:06 PM

    Thanks Colin, this is great information.


    Definitely moving them into the rooms would help but I am not sure if it is feasible unfortunately.

    The plan is to move them out of the ceiling and attach them to the T-bars (exposing them).
    They also seem to be positioned in line with the cinder block walls, therefore moving them slightly like below would probably help. (rooms 143,144,146 would probably suffer -- green dot signifies a potential new AP)



    Moving them into the rooms would be ideal and as the VRD pointed out, since the building has high attenuation walls/floors between rooms and between floors, microcells can come in handy.

    Purchase of more APs and moving them to the 2nd floor with low transmit power can possibly solve the issue most likely only if budget allows unfortunately and if testing proves so.


    What more detail is needed as I can provide more if needed..

  • 4.  RE: Improve this terrible network design

    Posted Nov 08, 2013 04:10 PM

    You probably will not have to lower the transmit power at all, because the cinder block will provide natural attenuation that you will need to keep them from interfering with each other.  Because of the cinder block you might be able to run them close to full power, maximizing your coverage.  You can probably do nothing until you first put an AP in a room on the 1st floor and measure left-right-above-below to see the coverage.  Do the same thing with one access point on the second floor.


  • 5.  RE: Improve this terrible network design

    Posted Nov 08, 2013 04:16 PM
    That's what I was thinking. We have someone going there Tuesday morning.
    2nd Floor has no APs but we are allowed to run a temporary cable for the purpose of this exercise.

    What do you think about all of these being RAPs?

  • 6.  RE: Improve this terrible network design

    Posted Nov 08, 2013 04:17 PM

    @pmonardo wrote:
    That's what I was thinking. We have someone going there Tuesday morning.
    2nd Floor has no APs but we are allowed to run a temporary cable for the purpose of this exercise.

    What do you think about all of these being RAPs?

    There should be no problem that these are RAPs, based on the limited information in this diagram.

  • 7.  RE: Improve this terrible network design

    Posted Nov 08, 2013 04:22 PM
    The only issue I see with RAPs is ARM not working between APs to best choose TX and channels.

    I have some APs with the same channels next to each other, e.g. AP7 and 2 (rooms 118-122). Client aware is enabled as well.

  • 8.  RE: Improve this terrible network design

    Posted Nov 08, 2013 04:24 PM

    ARM on RAPs should not function any differently as long as they terminate on the same controller.  If the AP is disconnected from the controller, channels and power will not change, of course.  It is quite possible that you have other RF that might be interfering.  Since these are AP93s, you have only 3 channels, but you will have more attenuation if they are in the rooms as opposed to if they are in the halls.  That will make two access points next to each other having the same channel less of an issue.

  • 9.  RE: Improve this terrible network design

    Posted Nov 08, 2013 04:26 PM

    Interesting, was told otherwise.

    The interesting thing is that there are barely any other wifi signals around the hotel.

    Hopefully we will be able to move the APs to the rooms to test. More to come..

  • 10.  RE: Improve this terrible network design

    Posted Nov 08, 2013 06:08 PM

    We had a similar challenge this past summer in our dorms with that type of wall material in between rooms.

    A couple of things to take in consideration :


    - We tested placing the APs (105's or 135's) ceiling mounted (inside the room) but there was no 802.11a signal on the other side (-82dBm) with the highest power level and with the 802.11g we received a (-70 - 75 dBm) and the signal quality wasn't good.

    - Decided to place the APs wall mounted and noticed that the bleed between floors was much better and also would allow us to provide wireless coverage for 3 floors ( the room the AP is mounted plus the bottom and upper floors)

    - We spreaded the APs across different floors

    - ARM does a great job adjusting the power/channel changes but one of the things you have to take in consideration is that the 11g radios will hear each other too much because of the density of APs so you may want to lower the mix EIRP value to 6 and Max to 15 possibly even 12. We also added Air Monitors to help the ARM process 


    Screenshot 2013-11-08 16.48.04.png


    Screenshot 2013-11-08 18.57.37.png


    Looking at the RF pattern on the 93H attenas you could wall mounted them and get good signal in between floors but of course you should test this to make sure the floor-ceiling distance and materials in between floors doesn't cause too much attenuation 


    You could also do a virtual survey using Airwave Visual RF or the Standalone version and that also will give you an idea of how to approach it

    Screenshot 2013-11-08 16.28.53.png

  • 11.  RE: Improve this terrible network design

    Posted Nov 11, 2013 07:57 AM
    Thanks Victor great feedback. I will see tomorrow how it goes. We are doing this during regular business hours (network will go up and down) as the hotel must remain in business of course.

    I'll check the antenna plots for the 93s as we do not use 93H's.

    I would love to put the AP's in the rooms permanently if that proves to be successful but will be a challenge to convince the hotel of that.

    Will let you all know tomorrow or wednesday how it goes.


  • 12.  RE: Improve this terrible network design

    Posted Nov 12, 2013 12:37 PM

    Not looking good.


    Found out APs in ceiling were facing upwards (93 logo facing ceiling). Moving APs out of ceiling and facing floor did not see to improve RSSI or SNR values on the 1st floor.


    Next step is to attempt to move 2 or 3 of them in a room staggered and test with it on the ceiling and wallmounted so we can test floor 1 and 2.


    What I am noticing is definitely roaming is not working properly (Probably because all RAPs), some sticky client issues. Even though APs are configured with min/max EIRP at 127 (some are not actually at MAX). 

    (c1) #show ap active  | include CN253
    CN253_AP6               group_YARCI               0            AP:HT:11/16/20       0                                 93       R2da   4h:19m:40s        
    CN253_AP7               group_YARCI               0            AP:HT:11/16/20       0                                 93       R2da   4h:19m:20s       
    CN253_AP9               group_YARCI               1            AP:HT:1/17/20        0                                 93       R2da   4h:16m:38s      
    CN253_AP2               group_YARCI               0            AP:HT:1/17/20        0                                 93       R2da   4h:16m:14s       
    CN253_AP3               group_YARCI               2            AP:HT:1/17/20        0                                 93       R2da   4h:15m:27s       
    CN253_AP4               group_YARCI               1            AP:HT:1/17/20        0                                 93       R2da   4h:15m:15s       
    CN253_AP1               group_YARCI             1            AP:HT:11/16/20       0                                 93       R2da   2h:12m:40s        
    CN253_AP8               group_YARCI             3            AP:HT:1/17/20        0                                 93       R2da   2h:2m:47s      
    CN253_AP10              group_YARCI            0            AP:HT:6/18/20        0                                 93       R2da   1h:15m:15s      
    CN253_AP5               group_YARCI               2            AP:HT:6/20/20        0                                 93       R2da   4h:20m:0s         

     AP 8 and AP10 were fighting for clients, made an AP specific profile for AP10 and reduced its power setting so that it users in the lobby would stop connecting to AP10. This did not seem to affect the rooms close to AP10.


    APs are definitely fighting for clients.

    I'll see once we can test in the rooms how it works out.


    EDIT: Moving an AP into the room definitely increased the SNR and RSSI within that room and the room next to it was not much better. Moving across the hall didn't seem to help and dropped the SNR by half.

    Placing the AP vertically along the wall and going upstairs to rooms 230/232 proved to me that placing the APs vertically in this building would not make a difference. The values we were seeing were similar to what we saw across the hall, SNR ~20 and RSSI -80s+


    Looks like I can conclude the following.

    1. Add more APs to the 2nd floor

    2. Even though the hotel may not want, add APs to the rooms.

    3. Because of #2, add way more APs.

    4. Possibly switch 93s for 105/115/135, an AP with a higher gain antenna.