A novel smartphone application for indoor positioning of users based on machine learning

Abstract

Smartphones are linked with individuals and are valuable and yet easily available sources for characterizing users' behavior and activities. User's location is among the characteristics of each individual that can be utilized in the provision of location-based services (LBs) in numerous scenarios such as remote health-care and interactive museums. Mobile phone tracking and positioning techniques approximate the position of a mobile phone and thereby its user, by disclosing the actual coordinate of a mobile phone. Considering the advances in positioning techniques, indoor positioning is still a challenging issue, because the coverage of satellite signals is limited in indoor environments. One of the promising solutions for indoor positioning is fingerprinting in which the signals of some known transmitters are measured in several reference points (RPs). This measured data, which is called dataset is stored and used to train a mathematical

Joint Coordinate Optimization in Fingerprint-Based Indoor Positioning

“…The proposed method has been implemented using the R programming language, and we use a dataset introduced in [5] to evaluate it over competitors. This dataset consists of 250 points, and each point has 75 samples from 27 Wi-Fi APs where the area size is 120 2 .…”

“…Here, 9 out of 27 APs that are more powerful in the environment have been selected. We conduct several experiments to compare the proposed 2D-GPR algorithm with three of the most popular baseline PRAs including CGPR [4], [5], CSVR [6], and CRF [7]. We use the Monte-Carlo cross-validation method [11] to evaluate these algorithms in which the train and test samples from 250 points are randomly selected times, and the average of test errors is reported to reduce the error bias.…”

“…It can be seen that when the number of RPs and APs is limited, 2D-GPR is more effective than other PRAs. The best record of 2D-GPR compared with other PRAs occurs in (40,5,15). It means that when there is one RP per 3 2 and one AP per 24 2 the proposed method is more effective.…”

“…The conventional PRAs have been created for single output scenarios, and this limitation forces us to utilize these algorithms for and coordinates separately. These algorithms, such as conventional Gaussian process regression (CGPR) [4], [5], support vector regression (CSVR) [6], and random forest (CRF) [7] have been used for positioning purposes, in which the optimization is performed for and coordinates separately. Some GPR-based algorithms have been developed for multi-task learning problems [8].…”

CSI-Based Probabilistic Indoor Position Determination: An Entropy Solution

“…Additionally, numerous literatures [19], [20], [33] further reveal such superior localization performance of the probabilistic algorithms over their deterministic rivals in the complex indoor environment. Generic probabilistic methods include the Bayesian network [34], Kullback-Leibler Divergence (KLD) [35], [36], Gaussian process [37], etc.. The essential cause resides in the fact that PDF contains the complete statistical characterizations of the complex random variables, which are capable of providing better location-specific RF signatures.…”