A device free high-precision indoor positioning and tracking method based on GMM-WKNN algorithm.

To address the high deployment complexity and algorithmic intricacies associated with current indoor target localization and tracking methods, this paper presents a Wi-Fi CSI indoor localization and tracking algorithm that integrates a Gaussian Mixture Model (GMM) with Weighted K-Nearest Neighbors (WKNN) and Kalman filtering. Initially, offline fingerprint information is collected from the indoor environment to establish an offline fingerprint database using the GMM. During the online phase, the trajectory information of the target individuals is gathered, and the clustering capabilities of the GMM are employed to optimize the grouping of Channel State Information (CSI) data. By categorizing the CSI data into distinct groups and assigning appropriate k-values for each group, we then perform initial trajectory estimation using the WKNN algorithm. Finally, the trajectory estimation is refined through a Kalman filter tracking model, achieving effective passive tracking of individuals indoors. In indoor environments, GMM effectively captures complex channel characteristics compared to other localization and tracking algorithms, demonstrating significant filtering and noise reduction capabilities. Experimental results demonstrate that the proposed localization algorithm significantly improves tracking accuracy compared to traditional localization methods and the CNN-based approach.