Indoor Localization Using Bluetooth-LE Beacons

Persons and devices in indoor environments, such as office buildings, may determine their location using Bluetooth LE beacons, such as iBeacons. Some number of these beacons are distributed over the environment of interest and their identifiers and locations are broadcast widely. The vector of received signal strengths from all these beacons may be intuitively expected to correlate well with location in the physical environment. However, the complexities of Bluetooth signal propagation in environments with obstructions and channels (walls, furniture, ducts, etc.) make it difficult to compute locations in this manner from only the signal values and known locations of beacons. Instead, a data-driven approach that uses a training set composed of observed signal strength vectors at known locations is more effective. This paper studies such methods using a publicly available dataset obtained by collecting training data in an academic building.