DuTrack: Long-Term Indoor Human Tracking with Dual-Channel Sensing and Inference

TL;DR

DuTrack addresses long‑term indoor tracking drift by fusing Wi‑Fi velocity cues with absolute acoustic hyperbolic constraints from cross‑chirp TDoF signals. The system formulates a weighted, two‑modality optimization that couples velocity residuals with geometric hyperbola constraints, solved in real time with a Gauss‑Newton update. A synthetic data‑driven training pipeline supports initial position inference and model calibration, enabling robust operation without manual initialization. Experimental results on commodity hardware show an average localization error of $0.78\,\mathrm{m}$ and substantial improvements over purely model‑based or data‑driven baselines, highlighting the practicality and impact of multimodal sensing for reliable long‑term tracking in homes.

Abstract

Wi-Fi tracking technology demonstrates promising potential for future smart home and intelligent family care. Currently, accurate Wi-Fi tracking methods rely primarily on fine-grained velocity features. However, such velocity-based approaches suffer from the problem of accumulative errors, making it challenging to stably track users' trajectories over a long period of time. This paper presents DuTrack, a fusion-based tracking system for stable human tracking. The fundamental idea is to leverage the ubiquitous acoustic signals in households to rectify the accumulative Wi-Fi tracking error. Theoretically, Wi-Fi sensing in line-of-sight (LoS) and non-line-of-sight (NLoS) scenarios can be modeled as elliptical Fresnel zones and hyperbolic zones, respectively. By designing acoustic sensing signals, we are able to model the acoustic sensing zones as a series of hyperbolic clusters. We reveal how to fuse the fields of electromagnetic waves and mechanical waves, and establish the optimization equation. Next, we design a data-driven architecture to solve the aforementioned optimization equation. Experimental results show that the proposed multimodal tracking scheme exhibits superior performance. We achieve a 89.37% reduction in median tracking error compared to model-based methods and a 65.02% reduction compared to data-driven methods.

Figures (13)

• Figure 1: Motivation behind DuTrack. (a) Conceptual illustration of drift accumulation in long-term indoor WiFi-based tracking, where the estimated trajectory gradually diverges from the true path. (b) In comparison, DuTrack effectively eliminates drift over time, maintaining alignment with the actual trajectory. (c) and (d) present real-world experimental results in which a person walks four consecutive laps along a rectangular path (outlined by a dashed line). (c) shows the output of a model-driven method, exhibiting significant drift across laps, while (d) demonstrates DuTrack's ability to preserve spatial consistency and suppress drift.
• Figure 2: PLCR-based tracking.
• Figure 3: The principle of acoustic sensing.
• Figure 4: Illustration of velocity composition for multimodality tracking in different NLoS scenarios.
• Figure 5: Impact of initial position.