Driving, biking, and walking to gain a sense of Wi-Fi geographies: Paul Torrens wore out the patience of his friends and family, but gathered 500,000 Wi-F samples across a 12 sq km area of Salt Lake City, Utah, for his paper "Wi-Fi Geographies," published in the 1 March 2008 issue of Annals of the Association of American Geographers. (The paper can be downloaded for a fee, but may be available through local public or academic libraries, too.)

In an interview recently, Dr. Torrens, an assistant professor at Arizona State University in the School of Geographical Sciences, said that he used his extended family to help him gather the data necessary to draw real conclusions. "Any time they were going anywhere, I got them to stick the rig to their car."

Dr. Torrens said that he decided to attack Wi-Fi because it was hard in the geographic field to find a subject area that hadn't been throughly explored, and that his interest in patterns and process over a landscape led him to Wi-Fi. His exploration looked at Wi-Fi as a topology overlaying population, demographics, and architecture.

In examining the literature to see if scholarly research had been carried out, he found a lot of wardriving details, but not a lot of accuracy or analysis. The maps of Wi-Fi coverage that are out there "rely on people going out and wardriving and submitting the data to some sort of online repository," Dr. Torrens said. While they may use GPS for timestamping and logging samples, "Unless you really know what you're doing with it, it provides very weak spatial accuracies [and] positional accuracies."

Dr. Torrens said, "I was able to come up with a much better accuracy." Some of his work is patented, and he said that while the university assembled the materials to file against his work, he remained a bit quiet about it. (As with most universities these days, ASU actively seeks to patent and license research as one means of funding the university's future.)

The data that he found in wardriving databases didn't account for quality, very few samples had timestamps, and where he found huge clusters, it didn't account for the timeframe, and thus was hard to tell whether the clusters existed at the same point in time. Dr. Torrens was collecting his data in 2005; wardriving databases may have improved in that time.

Dr. Torrens said that using techniques from the field, he could associate samples together, determining whether a cluster was legitimately such, or an abberration in the data--"whether a cluster is a cluster," in other words.

The research revealed some expected results, such as an extremely high number of access points in the most densely inhabited parts of town, but Dr. Torrens said he didn't expect to find that less-populated parts of town would also have a nearly ubiquitous spread of nodes. One area "that's relatively underpopulated is a whole warehouse district," he said, and they had lots of access points.

In the least-covered areas of the city, about seven access points were "visible"; in some places, that number was as high as 43 access points.

Also interesting to note was that security was most frequently enabled on Wi-Fi nodes in the parts of town dominated by students, who obviously had the technical jobs and understanding to prevent others from gaining access to their networks.

Dr. Torrens may carry out more Wi-Fi related geographic research, but that partly depends on having the resources or capability to gather information on a large scale. He'd love to gather live data that would allow him to show patterns as they change across the time of day or over a period of time.

"What I would like to do is to look at a temporal snapshot of the city, to look at how the Wi-Fi cloud is changing over time, over the course of a week," he said. "What is the temporal topography, the space-time topography of a city."

"To collect this kind of data set in real time would require a couple hundred thousand people with iPhones, citizen volunteers," he noted, but that might be possible with the capabilities of an iPhone software toolkit, promised by Apple in June, or through data sets gathered by firms like Skyhook Wireless.