A Profile-Matching Method for Wireless Positioning

This letter presents a profile-based wireless fingerprinting method, which mitigates positioning ambiguity by introducing short-term historical trajectories from dead-reckoning. This approach extends the fingerprint from one dimension to multiple, and thus enriches its diversity. Meanwhile, the multi-dimensional dynamic time warping method is introduced for matching multi-dimensional fingerprints with inaccurate profile lengths. Compared with the traditional single-point matching method, the profile-matching method provides a more reliable solution especially in environments with sparse distributed access points and a more accurate initial position. Indoor walking tests with two smartphones, in two buildings, and under four motion conditions illustrated that the proposed profile-matching method reduced the position errors by 11.5-21.6 %.