Research And Implementation Of Indoor Location Algorithm Based On Neural Network

Internet of things sensing technology‘s goal is to provide LBS(Location-based services),the key technology is how to positioning the object and person.The positioning for open-door places has mature technology like GPS navigator and cellular network location.But there is not much researches about indoor positioning,and there is no completed product on the market.This paper shows a way use both wifi signal and zigbee signal for positioning,wifi signal is for area position and zigbee signal is for coordinate position.The main work of this paper is as below:Shows an algorithm use kNN on wifi signal,used to get the location area of users.The GPS signal could not be used indoor,but in most cases there is a lot of wifi signals instead,they can be used for indoor positioning.In this design,build a dataset which contains the number of location and the wifi signal strength list of each location,when indoor positioning is needed,knn algorithm is used to compare the user's wifi signal strength around with the dataset and get the location number.When precise positioning is needed,zigbee signal should be used.There is two different methods for precise positioning in this paper,one is wrong judgment filter which is an improved algorithm of triangle centroid algorithm,the positioning accuracy is depend on the number of anchor points and the interval of each points.The other one is neural network method,this method could get stable result with only four anchor points.In the end of this paper,there is a compare of the methods mentioned in this paper :wifi finger print method,zigbee triangle centroid algorithm and neural network method.Besides,the advantages and disad-vantages of those methods are showed in a table.