Robust indoor localization based on hybrid Bayesian graphical models

In this paper, we study the problem of accurately estimating the location of indoor wireless devices which, due to its vast application areas, has continued to generate much interest both in the academia and industry. Specifically, we consider the location estimation problem and develop non-hybrid Bayesian location estimators for when received signal strength (RSS) and/or time of arrival (TOA) measurements between the target node (TN) and designated beacon nodes (BNs) in the network are obtainable. However, RSS- and TOA-based location estimators can be inaccurate when the RF-characteristics are not stable and time synchronization between the TN and BNs is not set up properly, respectively. To mitigate the disadvantages of these two location estimators, we propose the hybrid Bayesian location estimators. We show, through the results of extensive simulation experiments conducted that in comparison with the non-hybrid estimators, the hybrid Bayesian location estimators are more robust in the localization environment where RSS/TOA measurement precision varies.