Enhanced Radio Map Interpolation Methods Based on Dimensionality Reduction and Clustering

The received signal strength (RSS) based Wi-Fi fingerprinting method is one of the most potential and easily deployed approaches for a reliable indoor positioning system. However, due to labor intensive time-consuming radio map construction process, interpolation often incorporated. To ensure interpolated robust against environmental noise RSS fluctuations, we propose two novel methods, termed as DimRed DimRedClust, an improved construction. former performs dimensionality reduction prior while latter employs both clustering before interpolating map. For reduction, principal component analysis (PCA) or truncated singular value decomposition (TSVD) adopted profoundly extract essential features from data K-means algorithm used partition reference points (RPs) into several clusters. Subsequently, all virtual are via inverse distance weighting (IDW). Numerical results on real-world multi-floor multi-building dataset confirm supremacy proposed schemes over baseline IDW interpolation. Compared IDW, PCA-K-means-IDW, TSVD-K-means-IDW, PCA-IDW, TSVD-IDW could attain performance gain in terms average error up 30.17%, 30.93%, 19.33%, 21.61%, respectively.