Fuzzy rank cluster top k Euclidean distance and triangle based algorithm for magnetic field indoor positioning system

The indoor localisation based on magnetic field (MF) has drawn much research attention since they have a range of applications in science and industry. position estimation is generally the Euclidean distance (ED) between compared data points. Commonly, state-of-the-art k-nearest neighbour (KNN) algorithm used to estimate test point (TP) by considering average location closest estimated K reference points (RPs). However, problem using KNN fixed value does not guarantee accurate at every position. In this study, we first optimise MF RPs database clustering method. Each trained RP other nearby are clustered together certain distance. Then, create rank cluster where match top 10 ranks with nearest TP cluster. For proposed fuzzy algorithm, condition applied choose whether triangle area or find final Experiments show accuracy 5.88 m, which better than an improvement 31 %.