Research on Indoor Positioning Algorithm Based on Neighborhood Partitioning

In order to solve the problem of long matching time caused by a large number of reference fingerprints and inaccurate positioning caused by KNN algorithm alone. A method based on KNN partitioning is proposed in this paper. Firstly, the fingerprint space is clustered, and then the K nodes closest to the unknown node are obtained in the clustered area according to KNN algorithm. Secondly, the maximum and minimum coordinates of K fingerprints are used to determine the region, and then the region is divided by Newton interpolation method to form a virtual fingerprint matrix. Finally, KNN algorithm is used to re-determine the region, and then particle swarm optimization algorithm is used to find the optimal location node in this region iteratively. Experiments show that the algorithm can improve the positioning accuracy and reduce the matching time effectively, especially when the reference fingerprints are sparse.