The Effect of Density and Placement of BLE Beacons on Indoor Location and Motion Direction Estimation Accuracy

We investigated a method for simultaneously estimating the position and direction of a pedestrian walking indoors. The estimation is achieved using RSSIs (Received Signal Strength Indicators) from Bluetooth Low Energy (BLE) beacons detected by smartphones carried by pedestrians moving in one of eight directions. We investigated a method to estimate the position and moving direction of pedestrians using multiple time instances of RSSIs from multiple BLE beacons detected. The method was based on DNN (Deep Neural Network), which were trained to estimate the position and moving direction from a set of measured RSSIs. We fixed the size of the room to be estimated and compared the estimation accuracy by varying the number of beacons and their placement patterns. We also attempted to reduce the estimation error by treating the direction of movement as a continuous value. As a result, the estimation accuracy improved as the number of beacons increased, and a difference of 60.9% in the position estimation error and 61.2 % in the moving direction estimation error was seen when 4 and 8 beacons were used. Regarding the difference in the installation interval of beacons, we found that it is important to install them sparsely and evenly in the room.