Policy for Wi-Fi developers
There’s general agreement that the first step to happy coexistence between the various users of the RF commons is a set of policies for developers to follow, to separate their work from the corporate WLAN and each other. Ingredients can include:
- Channelization. One way is to keep channels 1, 44 and 48 (or whichever you prefer) for the corporate WLAN and insist that developers stay clear. It might be feasible to assign different channel sets to different development groups, but in Aruba at least, our cats won’t be herded that closely. We find one channel in the 2.4GHz band is enough to run a decent corporate WLAN these days, partly because all the PCs are 5GHz-capable and we use band-steering to keep them there. But many user devices are still on 2.4GHz.
- Alternatively, if test-AP signals are kept weak and sparse (see below) it may be possible to allow developers to use all 2.4GHz channels – but it is best if they only have channels 1, 6, 11 at 20MHz (rather than, say, channel 4 or 7, or a 40MHz channel), that way they don’t create noise across multiple channels.
- Transmit power. At maximum power, your transmissions will be heard clear across the floor – pretty much the whole building is a co-channel interference zone. It doesn’t take much effort to dial back the power when setting up a cubicle-scale lab. Policy can set transmit power limit for developers (e.g. 19dBm in both bands).
- Developers should only have their APs transmitting while testing is under way. This is easy to say, but a difficult behavior pattern to establish. If outreach is working (see below), switching off APs when not in use will become a habit.
- Identification. There are some nifty features in WLANs these days to match the wired and wireless sides of APs, but we consider it good practice for developers to put their names in SSIDs. That way we can quickly identify anyone who may have inadvertently transgressed.
Which brings us to cultural issues, including reminders, reinforcement and enforcement. Policies don’t work without outreach, so repetition on email, wikis, brown-bag lunches on Wi-Fi engineering, and having managers refer to Wi-Fi policy in their communication are all important.
Following that, it seems there is no substitute for one-on-one interaction: network engineers need to spend some of their work-week (one Aruba customer stipulates 6 – 8 person-hours per week as part of the job description) chasing down rogue APs, sitting with the owners to explain the need for RF cleanliness and helping to reconfigure the equipment. And when they have finished, leaving with a reminder that the airwaves are continually monitored. We also know of one Aruba customer who has automatic notification emails to those responsible for interfering APs, with escalation up the management chain for non-compliance.
It goes without saying that management support is critical, and people need to realize that these are basic precautions to make the air safe and clear for everyone, and it’s anti-social and bad for the business to be undisciplined.
Mitigating RF Propagation
Even with the most cooperative engineers, a relatively dense lab environment will raise the noise floor and cause co-channel interference for considerable distances. So in addition to segregating the IT WLAN from developers, it is also important to attenuate RF signals across the building as far as possible. The techniques to do this are well-known, here’s some recent experience:
- It is well-known that RF cages or shielded rooms are very effective. But they can also be expensive, and they tend to be small and hot and not so pleasant to work in. We use many shielded rooms in Sunnyvale, but they are not all the same. We need the most expensive anechoic chambers for RF measurements on antennas and new APs, but for most purposes it is possible to use simpler, less expensive shielded rooms. There’s quite a range available now, and it is worth looking into. These simpler rooms are quiet enough for Wi-Fi performance tests, and also shield the rest of the building from the tests inside.
- Even when tests are not in shielded rooms, there is value in mitigating the propagation of RF across the building. RF-shielding paint and wallpaper have been used for these purposes, and we know of at least one customer where this has been successful – but the conclusion was that it required an inordinate amount of time and effort. Since RF leaks readily through cracks and around corners, and the wavelength of Wi-Fi is measured in centimeters, a very thorough job of sealing is required to make a worthwhile difference. This means taking care of floors as well as walls (depending on building construction), paying attention to grounding all panels and sealing doorways. But ceilings are perhaps the most difficult, as it is often impractical to make a continuous, crevice-free and fully-grounded enclosure or false-ceiling.
Self-service and peer-pressure
Here is a very interesting innovation we came across recently at an Aruba customer. Graphic displays in high-usage areas display the current state-of-the-Wi-Fi-airwaves for that area. This gives those responsible a view of how well they are doing. One can also envisage a floorplan display highlighting and naming noisy Wi-Fi labs and their owners – allowing peer pressure to augment organizational discipline. It takes only a little work to automatically extract this information from WLAN management platforms via APIs.
Keeping the air clean is a continuous challenge in buildings where Wi-Fi-based product development uses the same spectrum as the corporate WLAN. We believe that best practice today is built on segregating development traffic by RF channel if feasible, and by common-sense guidelines like keeping transmit power as low as possible and switching off equipment when not in use. Partial RF attenuation through the use of paint and wallpaper can be surprisingly effective in limiting Wi-Fi interference zones, but it takes considerable effort to achieve a worthwhile level of attenuation. The best way to protect performance tests from outside interference, and to stop those tests interfering with the world outside is to use shielded rooms, and there is a balance between price and attenuation level – fully-sealed rooms are not needed for every purpose.
Finally, these policies are only effective if engineers follow them. We reported various techniques to raise and maintain awareness of the need for RF discipline, as well as an option where users are given information on how well they are doing and peer-pressure can work its magic.
Share your ideas
Many WLAN managers face similar issues – how do you approach them and which methods have you found effective?
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.