Sensor Aiding of HSGPS Pedestrian Navigation

This thesis illustrates how a set of low-cost, self-contained MEMS sensors can be used to aid a High Sensitivity GPS (HSGPS) based pedestrian navigation system in signal-degraded environments. The HSGPS utilizes long dwell time technology and provides excellent signal availability in areas such as urban canyons, where conventional GPS receivers barely provide any position solutions. Large signal attenuation and degradation effects characteristic of downtown areas, along with a high probability of acquiring echo-only or cross-correlation signals through the HSGPS receiver, introduce significant errors in HSGPS measurements. These, in turn, cause very significant position and velocity solution errors. In this situation, it is very hard to estimate measurement errors based on common GPS signal quality characteristics, such as a carrier-to-noise ratio or satellite elevation. Inaccurate estimates of measured signals ultimately result in an unknown distribution of HSGPS solution errors, which present a challenge for the proper modeling of the covariance matrix of HSGPS updates for system integration. For pedestrian navigation applications, sensor data is mechanized in a pedestrian dead reckoning (PDR) mode, where the position is propagated through the detection of user steps. In this way, errors are proportional to the distance traveled, and not to the time. The analysis of the effects of errors introduced by the primary PDR system parameters, such as heading and step length on PDR position accuracy, is studied. A method for modeling PDR position error growth versus traveled distance for several stochastic step-length and heading drift error models is proposed. Based on the results of individual system performance analysis, a method to integrate the sensor and HSGPS data is proposed. This proposed method utilizes a