September 13, 2004, 10:36 AM — By now you probably know that I'm the guy who "doesn't like Bluetooth" and who "doesn't like wireless LAN site surveys." Hey, it's not that I don't like Bluetooth - I really do appreciate what it can do. I just don't find it appropriate for many of the missions that it's been hyped for, and I think it's due to be replaced shortly by higher-performance technologies. The people at Ericsson recently halted all of Bluetooth's current and future Bluetooth development, and they're the folks who invented it! And it's not that I don't like site surveys, it's just that - no, wait a minute; I really don't like site surveys!
If you've never done one, a site survey is used to get a rough idea of the coverage of a given access point (AP) in a given location. You temporarily install power to an AP, with no need for a network connection, and then walk around with a mobile PC running a site-survey application, usually provided by the WLAN manufacturer. This gives you a general idea of the radio of coverage of the AP in the proposed location, and you can use this data, repeated across multiple temporary AP installations, to get an idea of how many APs will be required and to determine where they should go.
It sounds simple, but it's by and large a waste of time. Site surveys give you a snapshot of the performance of the AP in a given location, but it ignores real-world traffic patterns, data loads, number of users, time-boundedness constraints (which of increasing importance as voice-over-Wi-Fi comes online), and interference, just to name a few. A site survey does indeed give an idea of the coverage you can expect, but coverage is no longer the name of the game. In the good old days, when APs cost $2,000 or more, optimizing for the minimal number of APs made sense, especially given the low data rates of applications back then. Today, though APs are cheap, but the labor to install them remains expensive (and may even be getting more expensive, depending upon your location in the world and local regulations regarding AP installation). If we could somehow lower the labor costs, we could deploy lots of APs and move from a philosophy of scarcity to one of abundance, and do this at lower cost. We thus could optimize for capacity in place of coverage alone, and thus build WLANs designed to meet real-world challenges and that improve both productivity and user satisfaction. This is important as WLANs evolve into the default network connection for many users.
So, how do we cut the labor costs involved in AP installation? Simple - we plug the APs into existing wired connections, the same ones previously used for computers and such at the desktop. "But," I hear you saying, "won't the radio propagation (and thus the coverage) be less than had we installed the APs in the ceiling?" Yes, it will.