Evaluation of Localization Algorithms for WLAN-Based Tracking to Support Facility Management Field Activities

Facility management activities often require the capability to track and guide field personnel during routine and corrective maintenance tasks in dense indoor environments and large facilities. For example, during a water leak, a facility maintenance employee might require guidance to the nearest valve in a mechanical room. Such guidance requires accurate localization and tracking of mobile maintenance personnel in the field. The objective of this research study is to evaluate various localization algorithms for WLAN-based tracking of maintenance personnel in terms of accuracy and precision. Accuracy has been defined as the ability of a localization approach to track a person within a certain distance and precision has been defined as the ability to reproduce the required accuracy over time. The research described in this paper builds on the previous work of the authors on static user localization and utilizes the same test bed for evaluating the performance of different algorithms that utilizes WLAN technology to support mobile personnel tracking. The main motivation behind using the same test bed, which is an actively utilized building in Pittsburgh, PA, is to have the same baseline to evaluate the performance of static user localization and mobile user tracking. WLAN technology has been selected as it achieved good results for stationary personnel localization in the previous research work (Taneja et al. 2010). The authors have evaluated deterministic and probabilistic algorithms based on the fingerprinting approach (Bahl and Padmanabhan 2000) for mobile personnel tracking. The reason behind selecting fingerprinting approach is that this approach does not require line-of-sight between localization technology transmitters and mobile receivers, which was identified as a requirement in the previous research work (Taneja et al. 2010). The fingerprinting approach is further augmented by adding several filtering methods (Fox et al. 2003) to evaluate the impact of incorporating the motion characteristics of humans and the layout of the indoor environment on the accuracy and precision of the implemented algorithms. Initial assessment of the results indicate that deterministic algorithms perform better than probabilistic approaches when fingerprint data is limited, and incorporating the motion characteristics of humans and the layout of the indoor environment by implementing filtering methods, increases the accuracy and precision of mobile personnel tracking.