Smartphone Mode Detection for Positioning using Inertial Sensor

Abstract

Indoor Positioning has been in the center of attention in trending research. To this end, various means have been applied, including WiFi, Radio Frequency Identification (RFID), Fingerprinting, and Pedestrian Dead Reckoning (PDR). Smartphones, as an efficacious remedy for PDR technique parameters, are a serviceable choice due to their vast use. This article is dedicated to identifying and classifying different smartphone carrying patterns in different motion positions. Hence, we go through two steps; First using Machine Learning (ML) and Artificial Neural Networks(ANN), we identify smartphone carrying modes during user motions with four users and one smartphone to detect the suitable algorithm with the highest accuracy. Novelty of this paper is using Weighted K-Nearest Neighbor (WKNN) and ensemble by Genetic Algorithm (GA) with optimal weight, having offered notable results in categorizing. Furthermore, we review the smartphone sensor calibration effects on accuracy obtained by categorizing using four users and two smartphones in two states, before and after calibration using ML and ANN. The outcome was, calibration with smartphone sensors helps to categorize accuracy.