A Novel Received Signal Strength Estimation Method for Wi-Fi Fingerprint-Based Indoor Positioning

Wi-Fi fingerprint-based indoor positioning technology has now been extensively studied. However, this technology commonly requires an offline site survey process for construction of the fingerprint database which is extremely time-consuming and labor-intensive, hampering its large-scale implementation. To handle this issue, this paper proposes a Wi-Fi received signal strength (RSS) estimation method based on Gaussian Process Regression (GPR) and Autoencoder. Specifically, this paper first leverages GPR to learn the noise characteristics of signals, and employs Autoencoder to reconstruct values of RSS. Then, by integrating the output of both models, an estimation of RSS can be obtained. Extensive experimental results show that the algorithm proposed in this paper can effectively improve the accuracy of the fingerprint database reconstruction, compared with the traditional estimation algorithms, like standard GPR and K Nearest Neighbors (KNN).