基于自适应距离ADWKNN室内定位算法OACSTPCD

In the weighted K-nearest neighbor(WKNN)position fingerprint indoor positioning algorithm based on Bluetooth received signal strength(RSS),signal fluctuations can lead to low positioning accuracy when using a single distance estimation stan-dard.This article proposes an ADWKNN localization algorithm based on adaptive distance to address this issue.In the offline stage,K-means clustering algorithm is used to partition the fingerprint database to reduce the amount of data queries and ensure the timeli-ness of localization.In the online positioning stage,the RSS signals collected at the location point are Kalman filtered to reduce the interference of random noise,and then the ADWKNN algorithm is used to calculate the standard deviation of Manhattan distance and Euclidean distance,to select the distance estimation method adaptively and to realise the dynamic change of K value.The experi-mental results show that the average positioning accuracy of the ADWKNN algorithm is 1.22 m,which is a significant improvement compared with the WKNN algorithm using a single distance of cosine distance,Manhattan distance,Euclidean distance and So-rensen distance.