Dynamic Indoor Location Determination: Mechanisms and Robustness Evaluation

Many indoor location determination systems have been developed lately due to burgeoning demands on location based autonomous systems, however, the performance of most systems is not well understood because conditions under which reported systems were evaluated could be very different. To this date, there are still not yet many valuable evaluation studies using real measurement data sets in recent literature. In this work, we examine a set of localization mechanisms and evaluate their performance robustness under various configuration settings and dynamics using two different building environments including one regular office building and one underground floor-plan. We study both the distance-based and the database-based systems that employ only received signal strength for location determination, and propose several performance enhancement methods that would improve the robustness of common indoor systems. We introduce a simple but effective radio distortion model to capture signal perturbations under uncontrolled environments where signal values could be distorted significantly. Using the model, we also evaluate two location searching algorithms including lateration and multidimensional scaling (MDS), we observed that MDS presented stable and smaller localization errors under most experimental settings. Of systems studied in this paper, strong experimental evidence indicated that the signal-location map method, a database-based system, outperforms the distance-based systems, especially when environment is dynamic or when radio signal values are distorted intentionally.