RSRP difference elimination and motion state classification for fingerprint-based cellular network positioning system

Abstract

In recent years, the widespread availability of wireless communication has stimulated research of positioning service for cellular networks. Based on the minimization of drive-test introduced in 3GPP R10, it not only can be used to collect radio measurements and associated location information for network performance assessment, but also be used to build radio map for the fingerprint-based cellular network positioning system. However, due to the UE diversity, reference signal receiving power (RSRP) difference seriously degrades the positioning performance. In addition, UE motion changes the fingerprint length, which leads to fingerprint misalignment. Therefore, in this paper we propose to use multi-dimensional scaling algorithm to eliminate UE RSRP difference. By analyzing the relative RSRP difference of each pair of UE, we can eliminate RSRP difference in both offline and online. We also use the pattern recognition theory to classify UE motion for fingerprint alignment. We separated RSRP into two groups according to the UE motion state. The two groups of fingerprints are classified separately in offline and online. We implemented the proposed method in a real urban area and evaluated its positioning performance. The experiment results indicated our method can achieve a better positioning performance in cellular network positioning system.