In this paper, we propose a Surface Correlation-based fingerprinting technology using LTE signals in urban areas. The conventional fingerprinting method generally compares timely measured signal strength data with a database for the localization. The use of instantaneous data is highly susceptible to the signal noise and distortion in areas with low signal strength, resulting in performance degradation due to increased ambiguity. The Surface Correlation method utilizes a spatial pattern of signal strength to provide a stable and accurate location information. It converts timely measured signal strength into spatial domain using Pedestrian Dead-Reckoning (PDR). It also generates the reference spatial pattern that is matched to the pedestrian`s walking path from the fingerprinting database. The fingerprinting database consists of a link-structure to efficiently generate the reference pattern. To verify the proposed method, we have surveyed the LTE environment and constructed a fingerprinting database in the old town of Barcelona. Through several tests, we have confirmed that the performance of the proposed method is maintained at a 5m error level. Comparing Performance with the conventional fingerprinting and Particle Filter, we proved that the proposed method not only improves accuracy and stability but also improves the availability of LTE signal for localization.

Keywords: PDR, fingerprinting, LTE, RSS, pedestrian