Research On Indoor Location Technology Based On BP Neural Network And Improved Centroid Algorithm

With the rapid development of WLAN(Wireless Local Area Network)technology and Internet of Things technology, people’s demand for wireless locationis increasing day by day. Because NLOS(Non Line-of-Sight),multipath and variedinterference exist in the indoor environment, the common cellular mobilecommunication network and GPS(Global Position System) can not be successfullyapplied in indoor environment. However indoor environment is quite important forhuman beings, people desperately need the application of the indoor locationproducts, these requirements spurred the research and concerns on the indoorpositioning algorithm.Fistly we outline the research background, present situation and significance ofthe indoor location technology in this paper, discuss the advantages, disadvantagesand application environment of the classic indoor locating technology and algorithm.After that we introduce the ZigBee wireless communication technology, thenresearch the radio propagation model, analyse the effect of the parameters A and n inthe radio propagation model on the distance calculation use RSSI(Received SignalStrength Indicator), and their dependence on different environment, then we discussseveral classic centroid location algorithm separately.According to the problem of low accuracy, weak universality and too muchdependence on experience of the traditional indoor positioning algorithm based onRSSI, according to lots of research and analysis of wireless signal propagationmodel and the traditional wireless indoor positioning algorithm, we proposed a newmethod using BP(Back Propagation) neural network to fit the distance-loss model,thus avoid inaccurately calculating the parameters A and n in the radio propagationmodel. The input value of the BP neural network model is RSSI, the output value isthe corresponding distance. After obtaining the distance, the more precise six-pointcentroid algorithm instead of the traditional three-point centroid algorithm isproposed to calculate the location of the blind node. Based on the former monitoringplatform, we program a totally new high-integrated PC monitoring software, withthe help of the new software, we can take out, analyse and store the position datafrom the serial port, a new function that display the history track of the blind node isdeveloped, which make the new software more perfect. Finally we build the ZigBee wireless network positioning platform, then weconduct a experiment in the underground park of the school Renwen Building. Wethoroughly test and verify the newly proposed algorithm through the experiment,compared with the traditional positioning algorithm, the actual experimental resultsshow that the algorithm proposed reduces the positioning error, avoid fitting thecomplex parameters in the signal propagation model, improve the positioningaccuracy, prove the superiority of the proposed algorithm.