Improved Clustering Algorithm of Neighboring Reference Points Based on KNN for Indoor Localization

摘要

Reference points (RP) clustering methods such as K-means are frequently used to reduce the region of search in most of fingerprint clustering algorithms. However, traditional clustering algorithms analysis the geometric proximity of RP only in the off line phase, which has nothing to do with the test point. Meanwhile, both the clustering pattern and the number of clusters need to be predefined directly or indirectly, which means an unsuitable clustering pattern or an unsuitable number of clusters would lead to poor estimation accuracy. In this letter, in order to improve the performance of KNN algorithm with the neighboring RP selection, we utilize the k-means clustering algorithm to analysis the geometric proximity between RP and test point in the online phase. In the proposed algorithm, k-means clustering algorithm groups M(M<;K) nearest neighboring RPs according to their real physical distances to the test point instead of their signal distances. Then these M nearest neighboring RPs are used to estimate the location of the test point. Experiments were conducted in the fourteenth floor within an office building and the results demonstrate that the proposed method considerably outperforms the KNN algorithms in terms of positioning accuracy.