Efficient recalibration via Dynamic Matrix Completion

Fingerprint-based localization techniques have witnessed significant progress as they provide highly accurate location estimation with minimal hardware interventions. However, the required calibration phase is time and labour consuming. In this work, we propose a reduced effort recalibration technique for fingerprint-based indoor positioning systems. Particularly, we reduce the number of received fingerprints by performing spatial sub-sampling. The recovery of the full map from partial measurements is formulated as an instance of a Dynamic Matrix Completion problem where we exploit the spatio-temporal correlations among the fingerprints. Analytical studies and simulations are provided to evaluate the performance of the proposed technique in terms of reconstruction and location error.