Precise Indoor Positioning Based on Acoustic Ranging in Smartphone

• Published in: IEEE transactions on instrumentation and measurement 2021, Vol.70, p.1-12
• Main Authors: Chen, Ruizhi, Li, Zheng, Ye, Feng, Guo, Guangyi, Xu, Shihao, Qian, Long, Liu, Zuoya, Huang, Lixiong
• Format: Article
• Language: English
• Subjects: Accuracy; Acoustic signal; Acoustics; Algorithms; Bluetooth; Commercial off-the-shelf technology; Cross correlation; data fusion; Distance measurement; Dynamic tests; Extended Kalman filter; Feature extraction; indoor positioning; Inertial platforms; Measurement units; Signal detection; Smart phones; smartphone application; Smartphones; Time difference of arrival; Trajectory; Trajectory analysis; Ultrawideband
• ISSN: 0018-9456; 1557-9662
• DOI: 10.1109/TIM.2021.3082269

Existing indoor positioning systems based on Wi-Fi or Bluetooth confront difficulties and problems, such as low positioning accuracy, unstable, and small-signal coverage. A system using an acoustic ranging signal is presented in this article, and it is available for almost all commercial off-the-shelf (COTS) smartphones. A solution was developed to detect the modulated signals transmitted by specially designed anchor stations at each consisting of a speaker. The detector is a two-stage algorithm based on a generalized cross correlation (GCC) method and a time-frequency-based feature extraction. The detected timestamps are converted to time-difference-of-arrivals (TDOAs) observables, which can be applied with a least-squares (LS) estimator to obtain the smartphone positions in static cases. Furthermore, an extended Kalman filter (EKF) is introduced to integrate data from inertial measurement units (IMUs) and acoustic TDOA ranging observables for estimating a better trajectory in dynamic cases. Experimental results show that the signal detection algorithm is accurate with a mean error of about 0.3-1 m in terms of ranging estimation. The proposed method offers a static positioning accuracy of better than 0.5 m. For dynamic tests, positioning accuracy of 0.5 m at 50% and 1.6 m at 95% is achieved with four anchors in the first scene. In the second scene with a large test site, it achieves a dynamic positioning accuracy of 0.5 m at 50% and 1.82 m at 95% with eight anchors, which is comparable with the positioning accuracy of the ultrawideband (UWB) technology.