Wi-Fi Radio Map Interpolation with Sparse and Correlated Received Signal Strength Measurements for Indoor Positioning

Abstract

Wi-Fi based positioning fingerprint offers an accurate solution for indoor positioning techniques. It estimates the coordinates of a user or object by consulting an offline Wi-Fi radio map and searching for the best match of the currently observed Wi-Fi received signal strength (RSS) measurements. The construction of an offline Wi-Fi radio map is a laborious task in a large indoor floor plan. The offline radio map needs frequent maintenance if the data get faulted or need update due to the changes in indoor surroundings. This paper studies the effect of spatial correlation in the densely collected Wi-Fi measurements to enhanced the positioning accuracy. The K-nearest neighbour (KNN) and inverse distance weight (IDW) algorithms were implemented to interpolate the incomplete Wi-Fi radio map. The interpolation error is analysed with and without the correlation in the RSS measurements over different sparsity parameters. It is shown that at some sparsity parameter, the interpolation error reduces by 54% when the correlation exists in the collected Wi-Fi measurements.