Abstract:

[Objective] With the development of communication technology and the widespread use of mobile devices, Measurement Report (MR) data has been widely applied in wireless network positioning, and the MR fingerprint matching method based on fingerprint database is a widely used and relatively accurate method. However, the computational accuracy of the position depends on the quality of the data records in the MR fingerprint database. There is an urgent need to clean, filter, and position correct the data when accumulating the positional fingerprint database in order to ensure the data quality of the fingerprint database. [Methods] Therefore, this paper proposes a dirty fingerprint rejection algorithm for MR data localization scenarios, which identifies anomalous fingerprints through the DBSCAN clustering algorithm, corrects the location of the base station, and establishes a high-quality fingerprint database in order to improve the accuracy of MR data localization. [Results] In this paper, simulation data are used to conduct simulation experiments in three scenarios: rural, urban, and suburban, and the proposed method is analyzed and evaluated. [Conclusion] The experimental results show that with the removal of abnormal fingerprints and update of the fingerprint database, the accuracy of MR data localization has been effectively improved, and the proposed algorithm has advantages over other algorithms in terms of fingerprint recognition.

Key words: wireless network localization, MR fingerprint, high quality fingerprints, fingerprint database positioning