BVIRE improved algorithm for indoor localization based on RFID and a linear regression model

Abstract

Traditional indoor location technologies such as infrared technology and ultrasonic technology are complex,expensive, or having unsatisfactory location accuracy. Radio frequency identification (RFID) technologies are verypopular in many areas since their costs are very low. The tag in such technologies acts as the transmitter, and theradio signal strength indicator (RSSI) information is measured at the reader. However, RSSI information suffers strictlyfrom the multipath circumstance and circumferential elements. Therefore, the localization accuracy of the boundarywill be affected severely. In order to solve this problem, we introduce the boundary virtual reference label (BVIRE)algorithm to well utilize RFID techniques for locating the tracking object, which inserts some virtual reference tags onthe boundary by establishing a linear regression model that eliminates unwanted tag information from the estimationprocedure. The positioning accuracy of the boundary tags and stability have been improved significantly, at least 78%,without adding extra reference tags or radio frequency interference. Also, the estimation errors of our improved BVIREare much smaller compared to the virtual reference label, location identification based on the dynamic active RFIDcalibration (LANDMARC), ultrawide band, RADAR, and PinPoint algorithms.