Principal Component Analysis for Fingerprint Positioning

Due to the universal deployment of wireless LANs and the demands of indoor location based services, Wi-Fi fingerprinting has been investigated recently for localization. Regarding the influence of environment factors on RSSI measurements, this paper first presents fingerprint quantization method to form a quantized fingerprint database via threshold comparison. Then, we implement principal component analysis on the quantized fingerprint and generate dimensionality reduction fingerprint database. Based on these fingerprint, target localization is implemented. Finally, a comprehensive set of simulations are presented. We study and compare the localization accuracy under different fingerprint databases.