Bluetooth indoor positioning and correction method based on matrix completion and compressed sensing

Based on the received signal strength indication of Bluetooth low energy, a new fingerprint positioning method is proposed, which does not need to collect too many data points to establish fingerprint database and can correct the positioning error in time. In the offline stage, the low-rank matrix completion theory is used to recover a fingerprint database with only a small number of collection points; in the online stage, after clustering and matching the fingerprint database, the localization problem is transformed into compressed sensing model, and the localization result is obtained by solving the optimization problem. Experiment shows that the proposed positioning method can effectively reduce about 40% workload of early data acquisition and reduce the positioning error to 0.7m. Compared with the traditional KNN matching algorithm, the complexity of the algorithm is reduced and the accuracy is improved. © 2020 Computer Society of the Republic of China. All rights reserved.