Probabilistic Error Modeling and Topology-Based Smoothing of Indoor Localization and Tracking Data, Based on the IEEE 802.15.4a Chirp Spread Spectrum Specification

Location awareness is a core capability in many context-aware computing platforms. Multiple existing systems either provide inadequate accuracy or require extensive calibration or preexisting measurements in order to be functional. This work presents an extensive study of indoor tracking based on the chirp spread spectrum (CSS) specification and an associated analytical framework that allows comparisons to be made between different deployments. CSS provides resilience to fading, while being rapidly deployable. Wireless CSS modules are used to provide time of arrival measurements, necessary to infer the coordinates of a mobile user through trilateration. CSS resilience is tested in four deployments: an indoor space where line of sight (LoS) conditions are always satisfied, an indoor site that includes concrete, nonreflective obstructions, an industrial space with metallic, reflective obstacles, and a Tunnel. Empirical data are discussed in conjunction with the geometric dilution of precision (GDoP) metric, which depends on the system's deployment topology. The probabilistic modeling of the normalized localization error provides insight into the underlying distribution and is utilized in the context of a novel topology-based smoothing technique. Results indicate that CSS can provide accurate tracking. The application of the smoothing algorithm, however, further reduces the normalized error by a considerable amount.