Research On Improved LANDMARC Indoor Positioning Algorithm Based On Fingerprint Database

In recent years,the science and technology is developed rapidly,the urbanization process is accelerated constantly,Moreover the people's demand for positioning is raise significantly,especially in commercial push,logistics management,search and rescue.Outdoors,the location information can be obtained through the positioning technology of GNSS(Global Navigation and Positioning System).However,due to the occlusion of various outdoor buildings,the multipath propagation effect occurs during signal propagation,which causes to inferior indoor positioning.According to the statistics,more than 80% of people's dairy activities are indoors.Therefore,it is significant to develop a high-precision,low-cost,stable and reliable indoor positioning technology.Among the current mainstream indoor positioning technologies,RFID technology has low cost,high speed response,non-contact,non-line-of-sight,high relative accuracy of positioning.Thus,it is widely used in indoor positioning.Yet,its positioning accuracy and system design still need to be improved further.At first,the mainstream indoor positioning technologies are summarized.The mainstream positioning algorithms based on RFID technology are analyzed and the advantages and disadvantages of various technologies from various aspects are compared.Then the LANDMARC positioning algorithm based on the classic RSSI value is analyzed in detail.The factors which affect positioning performance are quantitatively analyzed by simulation.According to the analysis results,the following improvements are made to the LANDMARC positioning algorithm: Firstly,the LANDMARC positioning algorithm and location fingerprint algorithm are combined to reduce the unsteady RSSI(Received Signal Strength Indication)value.Fingerprint database is applied to correct the LANDMARC positioning algorithm and bring more references.Secondly,the label selection method is improved to select the nearest neighbor reference labels for the KNN(Nearest Neighborhood)algorithm which always gets outlier labels.Thirdly,The method to correct the algorithm error of the unknown point position by the algorithm estimation error of the adjacent label is proposed.Fourthly,the VIRE positioning algorithm is also improved by the fingerprint database.More accurate virtual labels are established through a stable fingerprint database.The fingerprint matching method is changed and the amount of calculation is reduced.The experiments show that the LANDMARC positioning algorithm based on the fingerprint database and the VIRE system have significantly improved the positioning accuracy compared with the original positioning algorithm,and the system stability is also greatly improved.