With the rapid development of science and technology,the continuous improvement of internet performance,and the large-scale construction of urban infrastructure,the network has further deepened the connection between each of us in our daily lives.From the primitive Internet to today’s wireless local area network-Wi Fi,the network and location information play an indelible role in our daily lives.From this,it can be seen that indoor location information has become increasingly important,which has sparked researchers’ research on Location Based Service(LBS).In recent years,Wi Fi fingerprint indoor positioning technology based on Received Signal Strength(RSS)has attracted widespread attention,as it only requires a mobile device and network,does not require additional infrastructure or other hardware devices,and does not require the collection of physical location information of Access Points(APs).However,localization using only received signal strength fingerprint information is easily affected by external dynamic environments and device heterogeneity.Using Signal Strength Differences(SSD)and Hyperbolic Location Fingerprint(HLF)to locate the received signal strength fingerprint information is a powerful measure to overcome device heterogeneity.The existing fusion based methods do not fully utilize the mutual fusion function of multiple fingerprints,resulting in low positioning accuracy.In order to further improve the positioning accuracy,a Multi Fingerprint Multi Classifier and Two layer Fusion Weight(FCW)positioning model with multiple fingerprints and classifiers is proposed.This method first collects signal strength difference fingerprints and hyperbolic position fingerprints from the received signal strength fingerprints,creates multiple fingerprint groups,and then selects three machine learning algorithms,It includes three classifiers:random forest(RF),K-Nearest Neighbor(KNN)and Support Vector Machine(SVM)to achieve indoor location.Then,a joint optimization algorithm with multi constraint twolayer fusion weights is proposed.This algorithm makes full use of the complementarity of multi fingerprint training classifiers,and then selects the best fusion weights to minimize the positioning error.In addition,a positioning method based on the fusion of dead reckoning algorithm and fingerprint technology has been proposed.The dead reckoning algorithm is used to estimate the relative position of reference points,and the original navigation technology is closely combined with the fingerprint.A classifier is selected to match and locate,which greatly enhances the dead reckoning technology and further improves the positioning accuracy. |