It seems that the more one pays attention to wireless site surveys, the more one realizes how limited the tools really are. I sure love the pretty RF graphics that illustrate where I may have a hole in my coverage. The trouble is, there’s so much more than meets the eye - literally. For instance, in a previous blog I noted that non-802.11 interfering devices don’t show up in wireless survey tools and hence the need for RF spectrum analyzers.
So what’s the problem with viewing what we can within the confines of our WLAN surveys? For starters, one must plan for capacity, not just signal coverage.
I once heard that a office worker moves to a different physical location on average every 18 months. At first glance, that doesn’t sound so bad, but that’s some 55 employees a month for a modest 1,000 employee company or location. Thus, the clustering or capacity needs can change rapidly – plan for re-surveying every 6 months to stay current and consider the labor hours needed to conduct such surveys.
Capacity also means more than just overlapping coverage. Certain areas may require more density – the bull pen cubicles vs. the warehouse for instance. That means paying more attention to overlapping AP channels and transmission power - the higher AP density the larger the user population we can support, providing of course, we limit the cell area by power (keep it low) and transmission speed (keep it high) to minimize interference from signals that can reach across multiple coverage boundaries, even with “honeycomb” channel 1-6-11 layouts.
Capacity is also better by using pure 802.11g (no mixing with 802.11b), provided all access points and clients are "g" capable. A sudden appearance of an 802.11b rogue can ruin everything. We could go to 802.11a but 5 GHz signals simply do not get through physical obstacles as well as 2.4 GHz.
The nature of the application is also important to capacity planning. VoIP over WLAN could present a challenge as most VoIP handsets use 802.11b, with a few using 802.11a. Cost will probably ultimately drive the wireless handset market towards b. So much for g.
Contrast VoIP with file transfer. VoIP doesn’t require a whole lot of bandwidth for one conversation (as low as 10 Kbps including overhead) but it does require very consistent packet delivery. Then we have competing high-bandwidth drag & drop file transfers or FTP. Even email can suck down wireless connections – think moderate to large attachments. Thus QoS policies must come into play.
Also think about users that don’t switch their laptops to a wired connection when they come into the office. I do it only because my 100 Mbps switched Ethernet connection is consistently faster than my shared 802.11g channel.
Also consider the third dimension – floors above/below you. You're fine if you are the sole tenant, but bad if not. You can’t do active site surveys on access points that you don’t own and must rely on basic beacon-based RF surveys that don’t account for speed vs. coverage. Beacon frames are usually sent at 1 or 2 Mbps that also employ a different modulation technique than the higher speeds and subsequently have a much further reach than normal data transmissions closer to the AP.
Add to this whole mess newer wireless technology like wireless switches, antenna arrays, beam forming technology, auto channel and power selecting APs, and the up and coming MIMO technology. Perhaps the answer is to provide saturated coverage everywhere (within the power and speed adjacency constraints mentioned above) and use troubleshooting tools to keep your end users happy.