A Wi-Fi Indoor Positioning Method Based on an Integration of EMDT and WKNN

Abstract

In indoor positioning, signal fluctuation is one of the main factors affecting positioning accuracy. To solve this problem, a new method based on an integration of the empirical mode decomposition threshold smoothing method (EMDT) and improved weighted K nearest neighbor (WKNN), named EMDT-WKNN, is proposed in this paper. First, the nonlinear and non-stationary received signal strength indication (RSSI) sequences are constructed. Secondly, intrinsic mode functions (IMF) selection criteria based on energy analysis method and fluctuation coefficients is proposed. Thirdly, the EMDT method is employed to smooth the RSSI fluctuation. Finally, to further avoid the influence of RSSI fluctuation on the positioning accuracy, the deviated matching points are removed, and more precise combined weights are constructed by combining the geometric distance of the matching points and the Euclidean distance of fingerprints in the positioning method-WKNN. The experimental results show that, on an underground parking dataset, the positioning accuracy based on EMDT-WKNN can reach 1.73 m in the 75th percentile positioning error, which is 27.6% better than 2.39 m of the original RSSI positioning method.

Research Progress of Wireless Positioning Methods Based on RSSI

“…This method can reduce the storage overhead and simultaneously improve the positioning accuracy. In [52], an Empirical Mode Decomposition Threshold (EMDT) was proposed to solve the problem of RSSI signal fluctuation. The method first preprocesses the RSSI fingerprint data using the EMDT method and proposes an improved WKNN algorithm, which first obtains the K initial RPs by WKNN, then calculates the center coordinates of the K RPs and the Euclidean distance d ic from each coordinate to the center coordinate.…”

Outlier Detection Performance of a Modified Z-Score Method in Time-Series RSS Observation With Hybrid Scale Estimators

“…Due to the dynamic nature of the indoor environment, such as the presence and absence of crowds and furniture, as well as variations in temporal and ambient conditions, the signals from the wireless APs fluctuate, causing the RSS measurements to fluctuate, resulting in poor localization accuracy [3], [4]. To deal with the RSS fluctuations, several research works have suggested collecting several RSS measurement observations over a time period and finding the mean average of these observations using techniques such as the mean-averaging filter, moving average filter, median filter, or Kalman filter [5], [6], [7]. During the collection of the time-series RSS observations, it is possible to have RSS outliers [5].…”