Exploiting Environmental Information Using HsMMs for Smartphone User Tracking

The extensive deployment of wireless infrastructure provides alternative low-cost methods for location awareness of mobile phone users (MPUs) in indoor environments by processing the received signal strength (RSS) of the mobile phone. In such a signal-processing framework, hidden Markov models (HMMs) are often used to model the uncertainties of RSS data and incorporate environmental information into localization. Since hidden semi-Markov models (HsMMs) outperform HMMs in their ability to model state duration more flexibly, employing HsMMs for indoor user positioning is a promising research direction. In this aspect, a user's personal preference for staying in a particular area, and the functionality of certain areas, such as a dining room, as well as navigation landmarks, can be utilized in the HsMM to assist localization. This article proposes an online HsMM forward recursion (HsMM-FR) algorithm to incorporate this information for real-time smartphone user tracking. We apply the proposed HsMM-FR algorithm to simulated, synthesized, and real RSS datasets in typical indoor environments for validation.