Effect of Calibration Point Density on Indoor Positioning Accuracy: a Study Based on Wi-Fi Fingerprinting Method

Abstract

Indoor positioning refers to all methods used in areas(indoor) where GNSS signals are too weak or non-existent for position determination, using various signals (Signals of Opportunity) and various sensor data. The availability of these signals and sensors for general navigation use is an important factor in terms of cost and feasibility. Considering the diversity of smart mobile devices and the technologies they contain; it is clear that they are perfect candidates for this job. Signals of Opportunity (SoOP) are intended for purposes other than navigation and Wi-Fi is a great example for this. Since majority of mobile devices have built-in Wi-Fi hardware, many studies focused on Wi-Fi positioning. This study used the fingerprint approach, which is among the most successful methods of indoor positioning using this technology. The number of calibration points to be marked in the calibration phase, which is the first of the two stages of this method, affects both the position accuracy and the time and effort spent. In this study, location accuracy was studied using NN, KNN and WKNN algorithms on a radio map with low calibration point density and it was discovered that the NN method provides both simplicity and satisfactory results in all scenarios. It was determined that the mean errors were minimal at the 2-meter point density and better results were obtained with the weighted-KNN algorithm compared to the KNN.