The Bad Science of WLAN Design
It’s been a while since I’ve posted anything. I’ve been working on a whitepaper which has been ‘blocking’ my blog writing for some time. Now though I have finally finished it and have published it below, following a brief introduction…
Time for a Whitepaper
For a long a time the general approach to WLAN design has irked me as you might guess from previous posts. Cutting a long story short, this lead to a conversation where I agreed to write down and e-mail my thoughts on automatic radio management to the representative from Aruba Networks, and from there, it was suggested that I actually write a white paper on the subject.
It kinda morphed into a more general critique of the state of the Wireless Industry. Features like Aruba’s ARM are not the really problem; they’re more a symptom of a bigger issue. It’s this bigger issue which is the focus of this paper.
And so, about a year later, here it is:
The Bad Science of WLAN design Nov 2014
I would be very interested in any thoughts or feedback as to similar experiences, anything you’d challenge as just plain wrong and how we might help the industry move forward from here.
I generally agree with the sentiments of the white paper and the need for more RD design knowledge and training.
Cisco gives a lot more than -67 as a guideline for a voice deployment than you give them credit for. The -67 is specific to the 7925 or 7921 VOIP phone. The 7925 deployment guide (I know, they actual make a VOIP phone) also goes on to say, -67 at cell edge, 20-30% overlap, less than 50% channel utilization, -92 noise floor, 25 SNR, <1% packet loss and <100ms of latency. That's a pretty healthy list of requirements.
Don't worry, since you failed to cite your source on where you got that, I went ahead and found you the link: http://www.cisco.com/c/dam/en/us/td/docs/voice_ip_comm/cuipph/7925g/7_0/english/deployment/guide/7925dply.pdf
Also, if you google "Cisco voice wireless guidelines." You will find multiple guides for design, config, troubleshoot and None of them simply say -67. Granted there are a lot of dated docs out there, and may not reflect all the current best practices, but to dismiss Cisco as not giving good guidelines as to what they would consider "Voice Grade" might be construed as vendor bashing.
Hi Jake, thanks for the feedback. The -67 was a figure which i’ve been using for a long time based on an early Cisco recommendation. I agree that voice spec you quoted is a much more healthy list. I ‘ve always wondered how do you measure “Cell Overlap” – do Cisco publish a methodology?
The problem is, I wonder how many “Wi-Fi outfits” are actually deliver WLANs according to said guidelines? Certainly the amount of this knowledge filtering through to the Education vertical in the UK seems limited.
Measuring overlap is done by going to the site with a testing tool like Eckahau or AirMagnet and physically testing each AP location. It can be alot of work but the results more than makeup for it from the technical side and the financial side.
Specifically, what would you be looking for within Ekahau/AirMagnet?
The deployment guides are out there for sure. The problem is that nobody seems to understand how we design to ensure we meet these criteria. The lack of knowledge is not just through the education vertical, in fact there is a staggering lack of understanding in the UK about how 802.11 WiFi actually works. On a daily basis we have to talk to customers who already have large WiFi networks deployed, but don’t understand the fundamentals of the technology. We spend hours explaining why we need to do onsite surveys and why we can’t just put more APs up. I am pursuing a CWNP certification path myself and while vendor deployment guides are great, CWNP gives you a deeper understanding of how we come the requirements outlined in the guides and how we go about designing for them.
I take Jake’s point about the vendor bashing, but on the whole this paper highlights the issues and conversations I come up against on a daily basis.
Even in the (recently) updated Cisco voice design guide (updated to WLC 7.3 code from the very old 4.2 code-based doco), Cisco *still* talks about cell overlap as a percentage.
It really just takes performing one VoWiFi survey, knowing the cell overlap requirement and then thinking… hmmm, how do you work that out *in the real world*. Then you start engaging your mind on how to work it out in the real world!
Keith wrote a post some time ago that covers off the way to achieve this, applicable to pre and post surveys. His post doesn’t delve into all of the nitty gritty but is pretty much the only solid source of this information out there! –> http://www.wlanpros.com/the-fallacy-of-channel-overlap/