Access Point Selection for Indoor Positioning Using Fingerprint Distance Maximization

This article addresses the issue of access point (AP) selection for indoor positioning using wireless networks. A fingerprint distance maximization (FDM) method is proposed. The novelty of the method is the treatment of the distance between fingerprints, where the APs are evaluated based on how frequently they contribute to the largest difference term between fingerprints measured at different locations, instead of the overall distance between fingerprints, which is conventionally applied. This strategy is effective because it increases the number of distinctive fingerprints in the database by preserving distinguishing features, even if these features are dominant in only a small number of fingerprints. Performance analysis is carried out by comparing the FDM method with several competing techniques using experimental data from a well-known dataset. It is shown that the FDM method leads to the highest accuracy and lowest uncertainty. In addition, the FDM method can be combined with the principal component analysis (PCA) to further reduce the dimensionality of the problem. Results show that for a fixed accuracy, the FDM-PCA technique reduces the data storage requirement by 30% and the online execution time by 5.5% compared to its closest competitor. The FDM-PCA technique is thus suitable for applications where mobile devices with limited storage and capacity are used as measuring instruments.

Automated summary

This article proposes a fingerprint distance maximization (FDM) method for AP selection in indoor positioning using wireless networks. The method evaluates APs based on their contribution to the largest difference term between fingerprints measured at different locations. Performance analysis shows that the FDM method achieves the highest accuracy and lowest uncertainty, and when combined with PCA, it reduces data storage requirement and execution time compared to competitors.