Regional Propagation Model Based Fingerprinting Localization in Indoor Environments

In this paper, we present a novel indoor propagation model for use in a wireless LAN (WLAN) based fingerprinting localization system. This model is based on a derivation of the free space propagation model taking into account the penetration loss and geographic propagation factors in different regions of the indoor structure. The objective of the model is to reconstruct the collected sparse fingerprinting database, eliminating the workload in the offline data collection phase. In the online phase, weighted K-nearest neighbor (WKNN) is adopted to obtain the final location estimation. We have carried out various experiments in a real-world setup to show the performance improvement of the proposed algorithm. Results show that proposed propagation model obtains more accurate prediction of the received signal strength values than other traditional models, delivering a high localization accuracy of up to 1.4 meter.