Benchmarking of Quantile-based Indoor Fingerprinting Algorithm

A growing demand for the information about location of numerous devices in indoor and urban environments raises the need for indoor localization. Indoor localization is needed for various applications and services, and it is considered as one of the key enablers of the Future Internet concepts. One of the most promising approaches in indoor localization is fingerprinting using information from the WiFi infrastructure. We propose a new fingerprinting-based indoor localization algorithm that makes use of the RSSI values from WiFi beacon packets for estimating the location. Namely, for generating fingerprints our algorithm uses the quantiles of RSSI values from beacon packets transmitted from various WiFi access points in the premises. Furthermore, the proposed algorithm uses Pompeiu-Hausdorff distance for calculating the difference between training fingerprints and ones generated by user to be localized. In the evaluation of the performance of indoor localization algorithms usual claims are the geometrical and room level accuracy of the algorithm. Unfortunately, due to the poorly defined scenarios and localization methodologies, those results are mostly unrepeatable and incomparable with other benchmarks. We aim on experimentally comparing the performance of our fingerprinting algorithm with three well known alternatives. For objective comparison of different algorithms we use guidelines and directions given in a newly developed EVARILOS Benchmarking Handbook (EBH). Following the classification in this handbook we evaluate the localization accuracy of different algorithms in three scenarios, namely small and big office and big open space scenarios. Using the systematic approach for evaluation of indoor localization solutions proposed in EBH we demonstrate that our algorithm achieves similar or better performance results in comparison to other algorithms in terms of geometrical and room level accuracy in the office scenarios, while our solution significantly prevail in the open-space scenario.