LTE signal-aided GNSS/PDR System for Mix-pose Pedestrian Positioning in Urban Environments Using Smartphone

Smartphone-based positioning is highly promising due to its widespread availability. However, its reliance on global navigation satellite systems (GNSS) leads to accuracy degradation in complex environments such as urban canyons. While integrating GNSS with inertial sensor-based pedestrian dead reckoning (PDR) can improve performance, challenges such as cumulative positioning errors as well as reduced accuracy in step length and heading estimation due to mixed phone poses remain unresolved during GNSS outages. To address these issues, a Long-Term Evolution (LTE) signal-aided GNSS/PDR system is proposed to improve the applicability to complex environments and mixed poses. Firstly, based on the analysis of propagation characteristics of LTE signal, we establish the relationship between the reference signal received power (RSRP) variations and positions to extract position, step length parameter and heading observations. Additionally, theoretical accuracy of the position observations is derived and its influencing factors are analyzed, providing a basis for the design of subsequent availability assessment strategy. Secondly, multiple observations from GNSS are extracted. Thirdly, after compensating for PDR heading offset caused by pose switching, we employ an unscented Kalman filter (UKF) model to fuse PDR with multi-observation from GNSS and LTE signals. An availability assessment strategy is designed to adaptively adjust filter parameters to exclude moments when GNSS or LTE signals are unavailable, which can mitigate the step length and heading errors induced by environments and poses effectively. Experiments conducted at three sites demonstrate that the introduction of LTE signal significantly reduces step length and heading errors, thereby improving positioning performance.