New Reconstructed Database for Cost Reduction in Indoor Fingerprinting Localization

Abstract

Location fingerprinting is a technique widely suggested for challenging indoor positioning. Despite the significant benefits of this technique, it needs a considerable amount of time and energy to measure the Received Signal Strength (RSS) at Reference Points (RPs) and build a fingerprinting database to achieve an appropriate localization accuracy. Reducing the number of RPs can reduce this cost, but it noticeably degrades the accuracy of positioning. In order to alleviate this problem, this paper takes the interior architecture of the indoor area and signal propagation effects into account and proposes two novel recovery methods for creating the reconstructed database instead of the measured one. They only need a few numbers of RPs to reconstruct the database and even are able to produce a denser database. The first method is a new zone-based path-loss propagation model which employs fingerprints of different zones separately and the second one is a new interpolation method, zone-based Weighted Ring-based (WRB). The proposed methods are compared with the conventional path-loss model and six interpolation functions. Two different test environments along with a benchmarking testbed, and various RPs configurations are also utilized to verify the proposed recovery methods, based on the reconstruction errors and the localization accuracies they provide. The results indicate that by taking only 11% of the initial RPs, the new zone-based path-loss model decreases the localization error up to 26% compared to the conventional path-loss model and the proposed zone-based WRB method outperforms all the other interpolation methods and improves the accuracy by 40%.