A Method of Radio Map Construction based on Crowdsourcing and Interpolation for Wi-Fi Positioning System

Wi-Fi indoor positioning has attracted a great deal of attention in recent years. Received signal strength (RSS)-based fingerprinting localization method may be the simplest and most efficient one. The radio map construction is time-consuming and labor-sensitive, hindering the wide use of RSS-based fingerprinting localization. Therefore, we propose a method of radio map construction based on crowdsourcing and interpolation for Wi-Fi indoor positioning system. First of all, fingerprints at a small number of refer points are collected by multiple smartphones. All missing RSS values are replaced by a constant. To alleviate the effect of device heterogeneity, the normalization method is introduced to both offline and online stage. Initial radio map is built after the process of normalization. Then, interpolated radio map can be generated by inverse distance weighted method. Subsequently, the initial and interpolated radio map are combined into a new one. Moreover, the principal component analysis is used to reduce dimensions for decreasing computation. Several experiments are conducted to evaluate the performance of the proposed method by comparing with others. Experimental results show that the proposed method has consistent positioning accuracy of manual radio map with saving large amounts of time and storage.