Mobile Cellular Based Fingerprinting Approach For Indoor Positioning

With the development of wireless communication technology,location-based services and applications are increasingly popular,such as emergency rescue,social network and other fields.In outdoor environment,GPS can provide accurate positioning.But in indoor environment,GPS will become worse or even could not work affected by the shielding of buildings or signal multipath transmission.Most researches have focused on indoor localization problem in recent years.Till now,location fingerprint approach shows higher accuracy.This thesis studies mobile cellular based fingerprint localization algorithm,to improve positioning accuracy and floor determination accuracy.The achieved works are as follows:1.The location areas were divided into grid,the RSSI of sampling points were measured and the collected data were analyzed.2.In order to reduce the influence of noise interference,missing data and compression of original data to location matching,data filtering,data completion,exponential transformation and other pretreatments were completed,and fingerprint database was established.3.Based on Weighted K Nearest Neighbor(WKNN)algorithm,Euclidean distance,Mahalanobis distance and Chebyshev distance,etc.are used as similarity measurement criteria.The experiment results show that the preprocessing of data can reduce the positioning error.At the same time,the influence of K value in WKNN algorithm on positioning error is analyzed.An improved localization algorithm is proposed to assign different weights to fingerprint points according to the distance between the fingerprint points and the center points.It gets 12%improvements of positioning accuracy,as shown in simulation.4.This thesis proposed a base station matching WKNN algorithm to solve the position problem in three dimensions space.The experimental results show that the proposed algorithm improves the accuracy of floor determination by 10%,comparing to classic WKNN algorithm.