Research On RFID Indoor Location Algorithm Based On BP Neural Network

Internet of things technology is following the modern computer,mobile Internet technology,but also a great opportunity for the development of information industry in the world,it can realize people to people,people to objects,objects to objects in close contact and communicate with each other.Radio frequency identification(RFID)technology as an important part of the concept of Internet of things,the technology has been widely used in indoor positioning,vehicle identification,access security,logistics,security products and other fields.Indoor positioning can provide tracking,positioning and navigation services effective for the specified object and users in the indoor environment,parking lot,shopping malls,train stations and other places to locate and guide the increasingly strong demand,in addition,precision marketing,intelligent manufacturing,intelligent logistics and other industries also need computer system capable of real-time calculation,identification of specific objects position.These requirements provide great opportunities for indoor positioning systems(Indoor Localization System,ILS).In contrast to other indoor positioning technology,such as video analysis,ultrasonic,infrared technology,RFID wir eless indoor positioning technology has the characteristics of wide coverage,high precision and low cost.This article is based on the low cost,which has the function of data processing and communication of commercial nRF24LE1 chip,RFID equipment,combined with position fingerprint positioning technology,K-means algorithm,support ve ctor machine algorithm(support vector machine,SVM)and BP(Back Propagation)ne ural network algorithm,the design of indoor wireless location algorithm has higher positioning accuracy.Firstly,this paper summarizes the current status and research achievements of indoor positioning technology,then introduces the working principle of radio frequency identification system,then introduces the transmission characteristics of the nRF24LE1 chip RFID devices and data communication,Then discusses the K-means algorithm,the support vector machine algorithm and BP neural network algorithm,and Illustrate the role of the algorithmin this paper.Based on the principle of the RFID device can pr ovide 4 discrete radio frequency signal output power,which are-18 dBm,-12 dBm,-6d Bm and 0dBm,and the longest transmission distance respectively were 4.5m,5m,5.5m and 6m.According to the characters of the class,in this paper,deploymenta 4 m * 6 m location area,and 10 reference tags according to the interval of 2 m spacing placed in a location area such as the edge of the area,and the positioning area is divided into 96 length of 0.5 m square grid,collecting RF signal output power of 10 ref erencetags on each small square grid,96 groups of fingerprint data is established.Using K-means clustering algorithm to macro regional division in locate areas,establish ed the SVM classification model and BP neural network model in each macro area.finally through the simulation experiment to test the performance of localization algorith m in this paper,the results show that when the location area is divided into two 2,3,4 class,root mean square error of the algorithm were 1.08 M,0.92 M,0.95 M,visibly When is divided into three kinds of macro areas,it has the minimum error.Thr ough experiment simulation,comparing with the traditional BP neural network localization algorithm,it is concluded that this algorithm has the characteristics of high precision and easy implementation.It has certain reference significance in wireless indoor po sitioning technology research based on the communication of discrete power output chip.