Fast features reduction of radio maps for real-time fingerprint-based wireless positioning systems

In fingerprint-based wireless positioning, a high number of wireless access points solicits feature reduction to obtain a compact radio map for accurate real-time positioning. Although principal component analysis (PCA) can be used to reduce dimensionality, PCA is computationally expensive. Additionally, PCA maps the data to a new space where physical meaning of the original features is lost. Presented is a faster features reduction approach using fast orthogonal search which selects the most informative features in the original space. The algorithm is applied to select the most informative access points in a radio map for accurate real-time wireless positioning. Experiments demonstrate the proposed method’s superior performance to PCA in terms of speed and slightly better performance in terms of accuracy.