Indoor Localization using Bayesian Filtering
Mgr. Miroslav Opiela
Ústav informatiky, PF UPJŠ
The indoor localization problem is considered to be a time-sequential, non-linear and non-Gaussian state estimation problem. Bayesian filtering is the commonly used stochastic filtering techique for the indoor positioning, where the current state of a system is computed from its previous state based on observations (in this case obtained from the smartphone sensors).
Under some assumption, Kalman filter and grid-based method provide the optimal solution for the posterior probability density computation. In other cases, including indoor positioning, Kalman filter, particle filter and grid-based method only approximate the posterior density. We will explain these Bayes filters when applied on the indoor positioning problem and provide evaluation results for the grid-based approach focusing on the influence of using the discrete filter and the localization accuracy for different grid types and step length estimations.