An experimental approach helps researchers assess the accuracy of indoor location systems.

From the “Find my Friends” feature on your phone to the global positioning system (GPS) in your vehicle, we have come to rely on the location technologies. But imagine your mobile phone could tell you precisely which room in your home you left your car keys. Or, for business owners, help keep track of inventory. This functionality is possible with indoor location systems.

Locating Devices Indoors

University of Guelph engineering professor Dr. Petros Spachos [1] has used an experimental approach to assess the technology behind indoor localization systems. To calculate the position of an unknown device, a localization system requires real-time location information from devices with known positions. Outside, localization systems use GPS technology, but inside, GPS signals are not easily received. Varying room layouts, sizes, and obstacles can make accuracy even more challenging.

An Experimental Approach

Spachos and his collaborators, including U of G graduate student Sebastian Sadowski who is also a former U of G undergraduate student, explored the accuracy of indoor location techniques in the context of different wireless technologies and within different types of indoor environments. The researchers set up three scenarios. The first was a small meeting room with tables and chairs, representing low interference. The second, a high-interference environment, consisted of a small meeting room with tables, chairs, and transmitting devices. The third scenario was an average interference environment: a large computer lab with tables. The researchers compared the accuracy of two localization techniques in each scenario.

Petros Spachos (left) and U of G graduate student Sebastian Sadowski (right) presented this work at the 2019 Information Technology, Electronics and Mobile Communication Conference in Vancouver

The most accurate technique for all scenarios was the “k-nearest neighbour,” which is rooted in a method called “fingerprinting.” Fingerprinting creates a radio map of an area by measuring a received radio signal from several access points and storing that information in a database. An algorithm compares signal values from an unknown location with values from known locations in the database. The k-nearest neighbour algorithm calculates distances between all the points in the database and selects the nearest matches to estimate the receiver’s unknown location.

“Our dataset is available in an open-source format for future research,” says Spachos. “Further exploration could lead to improved indoor localization technologies with everyday practicality, just like we see outdoors with GPS and ‘Find my Friends’.”

Petros Spachos is an Assistant Professor in the School of Engineering.

This work was supported by an NSERC Discovery Grant.

Sadowski S, Spachos P, Plataniotis KN. Memoryless Techniques and Wireless Technologies for Indoor Localization with the Internet of Things [2]. IEEE Internet of Things Journal. 2020 May 5. doi: 10.1109/JIOT.2020.2992651.