10

The issue: CCI (Co-channel Interference) on 2.4GHz.

The Environment: 40 seats per classroom. Higher-ed, so extremely mixed client base. AP-on-a-stick set to 24Mb lowest Mandatory data rate & lowest Tx power setting (2 dBm)

The Problem: Needing to disable 2.4GHz radio's to achieve the required density does not seem optimal. Am I going about this wrong?

The plan:

I've used the following formula for AP density and number of radios:

  • TTc = Target TCP/UDP Throughput, per client
  • TCc = Throughput Capabilities of client
  • CUm = Maximum realistic Channel Utilization
  • c = Number of clients

    TTc / TCc / CUm * c = # of Radios

At 40 seats per classroom, assuming each student has a laptop and a smartphone or tablet, with a target throughput of 3Mbps for laptops & 1Mbps for smartphone/tablets:

  • Laptops: 3Mbps / 150 / .8 * 40 = 1
  • Smartphone/Tablet: 1Mbps / 30 / .8 * 40 = 1.6

Total radios needed per classroom = 2.6. Each AP having 2 radios, this brings me to 1 AP per classroom; if rounding down. All classrooms are in use, and fully populated every day. Current client base shows about 30% 5GHz capable, campus-wide. I expect this number to increase rapidly as most clients are becoming 5GHz capable.

Antenna in use: AIR-ANT2566P4W-R (Dual-band patch)

PDF picture is result of placing the AP/Antenna at the front of classroom (facing East, or away from the hallway). There is very little attenuation between the rooms (sheetrock walls). This is a 3-story building, with same layout on each floor.

Current plan:

One AP per classroom, with patch antenna mounted facing east, for classrooms which have an exterior wall (see diagram), except using Mechanical down-tilt on first floor to reduce the signal in the rooms 'behind' the Anetanna. 2nd & 3rd floors have antenna mounted as drawn, w/ no downtilt. Using 20MHz 5GHz channels, and disabling some of the 2.4GHz radio's, to reduce CCI.

Interior classrooms will utilize same patch antenna, again using down-tilt on first floor and not on upper floors. Disabling some of the 2.4GHz radio's. If down-tilt is utilized on 2nd/3rd floors, the signal penetrates the floors too much.

Options I've considered:

  • Purposefully creating attenuation 'behind' the antenna (mesh on wall, RF blocking paint etc.) to reduce the signal 'behind').
  • Using non-Cisco antenna, with a better front-to-back ratio, which would reduce CCI as a result of the radiation of signal 'behind'. Not sure if an equivelant one exists. Essentially, I'd benefit from a similiar radiation patter, minus the back-lobes.
  • Using internal antenna, and just disabling more 2.4GHz radio's (ea. antenna is ~500$)

The picture represents the results of having an antenna mounted in room 222, pointing out/east/towards the exterior wall; and client RSSI readings from various locations.

ap-on-a-stick_results

  • What type of antennae are you using and how do you plan to handle coverage when an AP for a room is out of service? – generalnetworkerror Sep 21 '13 at 9:22
  • @generalnetworkerror The antenna in use is linked in the description. Our AP's don't fail very often, but when it does happen, RRM (Radio Resource Management) detects this and increases TxPower of the surrounding AP's. – Mike A Sep 24 '13 at 16:46
6

You seem to be going about this the right way (which is the only real question I can find in your post), but I will throw in a few more considerations and/or options.

  1. With this many APs deployed and as many users as you expect (and the other non-802.11 devices they bring with such as bluetooth), what is the noise floor in 2.4GHz? It may be that your noise floor is in the -70 to -80 range, which means that "usable" signal outside the room is nominal.
  2. You mentioned RF paint (or mesh), behind the APs. Another option to consider is using it on all walls and deploying two APs per room (for redundancy and extra capacity). You may also need to use it on ceilings of rooms below. However, test any product before using it widely as some do not work as advertised.
  3. You mention that your transmit power is set to the lowest setting (2dBm) and your antenna is 6dBi. Since you can't turn down the transmit power anymore, you could consider a lower gain antennas or use attenuators to further decrease the signal strength. You will want to make sure that you are able to produce enough signal for a good connection (and high data rates) for a range of devices that will likely be found in the room.
  4. Test. Test. Test. High density deployments are a challenge, and the challenges change all the time as technology and the types of devices users bring into the environment are always changing. Unfortunately, you can't just set it in place and walk away.
  5. Realize that there will be problems and issues. This is just a fact of life in an environment where you don't have control over devices that will be brought onto your wireless network. Keep a list of device details that have been reported as having issues (type, make, wireless chipset, driver version, OS, what steps resolved the issue, etc) as this will allow you to start finding trends and general solutions so you can provide information (i.e. "try these steps" or "avoid this hardware") to your support staff and/or end users.
  6. Subscribe to mailing lists and utilize resources such as forums and industry blogs regularly to keep aware of problems that might be the result of OS updates or other problems. Some examples: Educause wireless-lan mailing list (probably applies since you are an education organization of some sort), Aruba Airheads forums, Ubiquiti forums, Cisco Wireless-Mobility support forums, Ruckus Room (while a number of these are vendor sites, there can be very good information on them and many wireless problems will apply to other vendor environments as well).

Good luck.

  • Is "TTc / TCc / CUm * c = # of Radios" a correct formula? – radio-free-europe Sep 27 '13 at 8:19
  • @radio-free-europe, not sold on the formula, first because at first look I think there is something off in it (which I haven't had the time to sit down with and play yet) and also because these short/simple/clean formulas are at best a quick guide. Wireless isn't that simple/clean (I was just at an edu where they installed a 120 node cluster to churn out algorithms used by predictive survey tools to calculate coverage/capacity). However, the bandwidth/number of clients in the example lines up with what most edu's I know are using for deployments (i.e. with those numbers, installing 1-2 APs). – YLearn Sep 27 '13 at 21:36
  • @YLearn Thanks, all good points. RF Attenuators seem favorable to "paint/mesh", however point 4 is the real answer. test, test, test. Not very easy to simulate the environment, but a combination of active site surveying & analyzing existing classroom environments RF (to identify noise floor) is the plan. – Mike A Sep 30 '13 at 22:46

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.