Research On BLE Indoor Fingerprint Location Algorithm Based On Cluster Analysis

With the advancement of science and technology,indoor positioning technology has developed rapidly,and the demand for people's location information has become more and more intense.Indoor positioning technology mainly includes radio frequency identification technology,Wi-Fi technology,infrared technology,ultra-wideband technology,Bluetooth technology,visible light technology,etc.The positioning method is divided into geometric measurement method,approximation method,location fingerprint method and so on.This article adopts the position fingerprint method that is more advantageous than other methods.The positioning technology selects the emerging low-power Bluetooth technology,which is low-cost,has a long life,and has lower power consumption.It is also simpler to deploy,easy to operate,and can provide long-lasting and stable signal transmission.The location fingerprint method is divided into two stages: offline database building stage and online positioning stage.The offline database building stage mainly builds a fingerprint database,also known as a radio map.The offline database building stage usually needs to use a clustering algorithm to perform clustering on the data set.Commonly used clustering algorithms include K-means algorithm.In this paper,the selection of initial points in the clustering algorithm is optimized.Through continuous binary clustering until the number of clusters meets the requirements,the centroid of each cluster is finally used as the initial point,and the number of clusters is determined scientifically by the elbow method.The online positioning stage is mainly to match the fingerprint information collected in real time with the data in the fingerprint database.In this paper,the WKNN algorithm used in the online positioning stage is optimized,and a new classification model based on RSS and a certain compensation mechanism is proposed to avoid positioning errors caused by classification errors.The derivation of physics is compared in the selection of weight values,Manhattan distance,Euclidean distance,,designed a method to adaptively select the weight value based on the data situation of the point to be measured.Finally,this paper collected the measured data and verified and compared the algorithm.The specific hardware equipment is the NRF52832 chip signal transmission and reception device with NORDIC built-in produced by Shenzhen Rui Dilai Technology.The chip supports BLE5.0,which meets Experimental needs.The experiment shows the influence of different parameters on the positioning result,and also proves that the algorithm proposed in this paper is superior to other traditional algorithms.