After a while, the novelty of the site survey wore off. Still, it was fun to work with clients doing their first site surveys and to watch them marvel at how a given link in plain sight only got three bars, while that distant back corner under the metal staircase got five. Ah, nonlinearities. What would we do without them?

A quick education in multipath never hurt anyone.

And then it hit me: Site surveys were entertaining, but were they necessary? As the cost of hardware fell dramatically following the availability of 802.11-based products, the fun turned into a waste of time-and money.

For those of you who have not had the pleasure, a site survey is designed to map the coverage of a given access point (AP) in a particular location in a specific physical environment. One installs an AP and then walks around with a notebook computer running the vendor's site survey software and noting signal strength. In this way, one gains a good knowledge of likely coverage and thus can optimize the placement-and number-of APs required.

When access points cost $2,000 and handled a whopping 2 Mbits/second, site surveys were essential. Today, though, optimizing for coverage is a false economy. APs sell for only a few hundred dollars; switched architectures will drive this cost even lower.

Since longer distances mean less throughput, the very thinking behind the site survey is today suspect. Why spend thousands walking around to determine where to put a $300 box? And since site surveys say nothing about capacity, which is what we really should be optimizing for-well, you get my drift.

So, the best approach today is to install access points in what seem like the right places, measure the results in terms of coverage and capacity, and then add a couple of additional Aps if necessary.

In fact, this iterative process of "measure and add" is probably the most cost-effective approach you can take.

Craig J. Mathias is Principal of the Farpoint Group (Ashland, Mass.).