The Site Survey Question Revisited

The last in my current series of tests of Wi-Fi tools was just published in Network World, this time on the subject of site surveys. I have for many years urged caution in committing resources to a site survey on the grounds that, as prices for WLAN systems have fallen and performance has so dramatically improved, spending a good deal of time and money on a large-scale planning exercise fundamentally centered on coverage rather than capacity might have less of a return than even a trial-and-error approach. I have to say here that the tools I tested, both predictive (which build simulation models) and analytical (which work off of actual measurements) are quite capable, powerful, and relatively easy to use once one spends a little time learning how to use their features. Some product combinations include both techniques, allowing real-world feedback to refine simulations, and providing very powerful tools indeed, and some tools even consider performance requirements. Regardless, I don't think this project caused me to radically re-think my general strategy as noted above - many venues really won't require a formal site survey of any form. I always recommend a pre-installation RF sweep using one of the many good spectral-analysis tools available, just to make sure there are no significant interferers nearby, and a post-installation walk-around survey in the case of VoFi apps is probably a good idea to make sure that coverage meets requirements. Other than that, monitor traffic and users for problems and add an AP or two where required. Still, take a look at the review - there's a lot of functionality there, with the potential to stimulate a lot of beneficial thought on the part of planners and installers.Thus the big question remains how to measure the ROI of any form of site survey. Many in the crop of tools are I examined are amazingly detailed and capable, but one should never invest without an idea of what one gets for one's money. So, do sophisticated site surveys and modeling tools have such a return?To be honest, I don't know. The Network World article is a product review, not a formal test, and we concentrated here on making WLAN planners, installers, and operations staff aware of what's possible with these tools. Note that the WLAN system vendors also provide, in most cases, site-survey tools and, in some cases, modeling tools as well. But we didn't do a test to compare the effort, cost, and results obtained with modeling and/or a physical survey with simply installing, letting the associated management software set things up, and then evaluating the results in terms of throughput, coverage, capacity, and, of course, cost, over time. We didn't compare the tools for relative or absolute accuracy. Such would be an enormous exercise, and several of these tests would have to be done, on a reasonably large scale, before we could draw any valid conclusions.Still, I found that the analytical site-survey tools, regardless of origin, work very well and have obvious utility, subject to resolution of the ROI question. The predictive modeling tools, on the other hand, must demonstrate that there is a meaningful ROI associated with their use in order to achieve broad penetration. I love the concept - I've been working on products of this type since my days over a decade ago as the Chair of Wireless LAN Research Labs, at Worcester Polytechnic Institute, where a team developed one of the first of these - PlaceTool. So I'm personally a believer here, again subject to resolution of the cost/ROI issue.But that brings up a key conclusion: I think site surveys are going to eventually be viewed as a key element of the ongoing, day-to-day WLAN management process, not as a one-time, pre-installation activity. It's necessary to know that coverage and capacity are acceptable, and when they are not via alerts and alarms, so a self-monitoring infrastructure seems like the only way to go. That, to me, anyway, is the future of these tools, whether supplied by the systems vendor or via third parties. And I'd also like to see integration with RTLS systems to track mobile users and devices participating in the survey, the use of same to auto-place APs, and, most importantly, the ability to turn any client into a sensor for site-survey and other functions. That, of course, will likely have to wait for further action from 802.11, although 802.11k and .11v should help.