Indoor positioning made more precise

Korean engineers say they’ve made a major breakthrough in high-precision indoor positioning that will make it faster and more accurate.

GPS doesn’t work well in indoor spaces or urban canyons, as it requires a clear view to communicate with satellites. It’s also only one-third as accurate in the vertical direction as it is in the horizontal, making it impossible to locate a person or object within tall buildings.

For indoor positioning, therefore, location-based service providers have mostly used a combination of GPS and wireless network systems such as WiFi, cellular connectivity, Ultra Wide Band (UWB), or Radio-frequency Identification (RFID).

For example, the WiFi Positioning System (WPS) collects both GPS and WiFi signals, and is used by many companies including Google and Apple to give clients location information services.

“WPS is helpful to a certain extent, but it is not sufficient because the technology needs GPS signals to tag the location of WiFi fingerprints collected from mobile devices,” says Professor Dong-Soo Han from KAIST.

“Therefore, even if you are surrounded by rich WiFi signals, they can be useless unless accompanied with GPS signals. Our research team tried to solve this problem and finally came up with a radio map that is created based on WiFi fingerprints only.”

WiFi fingerprints are a set of WiFi signals captured by a mobile device, along with the measurements of received WiFi signal strengths (RSSs) from access points surrounding the device. A WiFi radio map shows the RSSs of WiFi access points (APs) at different locations in a given environment. This means that each WiFi fingerprint on the radio map is connected to location information.

The KAIST team collected fingerprints from users’ smartphones every 30 minutes, and analyzed the characteristics of the collected fingerprints.

“We discovered that mobile devices such as cell phones are not necessarily on the move all the time, meaning that they have locations where they stay for a certain period of time on a regular basis,” says Han. “If you have a full-time job, then your phone, at least, has a fixed location of home and office.”

With the help of Google’s geocoding, Han converted each home and office address into geographic coordinates to obtain the location of the collected fingerprints.

His team then selected four areas in Korea – a mix of commercial and residential locations – collected 7,000 WiFi fingerprints at 400 access points in each area, and created a WiFi radio map. Tests showed that once the data collection rate rose above 50 percent, the average error was less than 10m.