Low-Power Positioning on Wearable: Integration of Snapshot GNSS and Neural PDR

Chin Lok Tsang, Shiyu Bai, Di Hai, Hoi-Wah Ng, Li-Ta Hsu

Peer Reviewed

Abstract:	Outdoor positioning is essential for wearable devices, enabling users to track activities such as running, hiking, and other location-dependent applications. However, conventional GNSS receivers consume substantial power due to their reliance on prolonged continuous closed-loop signal tracking, which poses significant challenges for battery-constrained wearables. Snapshot GNSS (SGNSS) addresses this limitation by capturing brief signal snapshots and performing open-loop processing, significantly reducing power consumption but providing only discrete position fixes. Pedestrian dead reckoning (PDR) offers an alternative approach with continuous, low-power positioning; however, it suffers from cumulative drift errors over time. While existing GNSS-PDR fusion solutions can correct this drift, they still require power-intensive continuous GNSS tracking, leaving a gap in achieving both continuous positioning and low power consumption simultaneously. To address these limitations, this work proposes a novel snapshot GNSS neural PDR (SGNSS-NPDR) fusion framework that combines the complementary strengths of both systems: PDR provides continuous position tracking while periodic SGNSS fixes correct accumulated drift, all while maintaining low power consumption. The proposed system employs L5-band-based SGNSS to enhance measurement accuracy through improved signal power and multipath mitigation, and adopts a neural inertial odometry approach that directly learns the relationship between IMU measurements and velocity to improve velocity estimation accuracy. Experimental results demonstrate that SGNSS-NPDR achieves a favorable balance between positioning accuracy and power consumption, attaining a CEP68 of 3.47 meters and a CEP95 of 6.38 meters with 10-second SGNSS update intervals. These findings provide device manufacturers with flexible design options to balance power consumption and positioning accuracy requirements for wearable products.
Published in:	Proceedings of the ION 2026 Pacific PNT Meeting April 13 - 16, 2026 Hilton Waikiki Beach Honolulu, Hawaii
Pages:	525 - 540
Cite this article:	Tsang, Chin Lok, Bai, Shiyu, Hai, Di, Ng, Hoi-Wah, Hsu, Li-Ta, "Low-Power Positioning on Wearable: Integration of Snapshot GNSS and Neural PDR," Proceedings of the ION 2026 Pacific PNT Meeting, Honolulu, Hawaii, April 2026, pp. 525-540. https://doi.org/10.33012/2026.20631
Full Paper:	ION Members/Non-Members: 1 Download Credit Sign In