Smartphone-Based Indoor Navigation Using Pdr And Magnetic Matching

You Li,Xiaoji Niu,Peng Zhang,Haiyu Lan,Yuan Zhuang,Naser El-Sheimy

2015-01-01

Abstract:This paper presents an integrated pedestrian dead-reckoning (PDR) / Magnetic matching (MM) navigation algorithm with off-the-shelf sensors in smartphones. This method maximizes the advantages and minimizes the disadvantages of each technology. As magnetic matching is a profile-matching method, we calculate the rough length of the measured profile using the steps detected by accelerometers. Because accurate and real-time steplength estimation is still an open issue for pedestrian navigation, we utilize the dynamic time warping (DTW) algorithm for matching with an inaccurate profile length. Additionally, we detect the roll and pitch angles using accelerometers in smartphones and extract the vertical and horizontal magnetic intensities, in order to increase the magnetic fingerprint dimension without extra sensors. Furthermore, we use magnetic gradient fingerprints, instead of magnetic intensity fingerprints, to mitigate the effect of the diversity and the run-to-run biases of MEMS magnetometers. Tests with different smartphones indicated that the matched MM results were accurate and had small fluctuations. Nevertheless, although MM is a profile-matching algorithm, it had mismatches (i.e., matches to an incorrect point that were more than 20 m away from the true value) on several occasions. The integration of PDR and MM has the potential to provide continuous and reliable navigation results in indoor environments having frequent magnetic interferences, and under natural device motion modes including handheld, at the ear, dangling with hand, and in a pocket.