Research And Improvement Of Indoor Location Algorithm Based On RSSI

With the development of the Internet of Things,the Location Based Service(LBS)in indoor environment has attracted much attention.Indoor location technology has been further developed as well,and become an important research direction in the field of location.By analyzing several typical indoor location technologies,we can get that the indoor location technology based on WLAN has been the most widely used because of its unique advantages such as the low cost of the hardware,the high accuracy of the location and less influences from the indoor environment.This paper mainly attempts to investigate and improve the fingerprint location algorithm based on signal strength(RSSI),aiming at enhancing the stability of the positioning and improving the accuracy of positioning.This paper analyzes the characteristics of RSSI and the source of the error,and optimizes the fingerprint in the aspect of data preprocessing to improve the stability of the fingerprint database.Moreover,several different kinds of methods are analysised,such as mean filter,median filter,Gauss filter,Kalman filter.The result points out their advantages and disadvantages.Considering the complexity and positioning error,smooth filtering method based on average is proposed,and the experiments show that the method has a smaller positioning error while ensuring low complexity.In addition,by studying several principles of commonly used RSSI-based fingerprint matching algorithms,this paper points out the shortcomings of those existing fingerprint matching algorithms.On the basis of WKNN algorithm,K-means clustering algorithm could solve the problem of the large computation to some extent.However,the K-means clustering algorithm has a high dependence on the initial clustering center of random selection,which will lead to an unstable accuracy of positioning.In order to solve this problem,a fingerprint localization algorithm based on simulated annealing and K-means clustering is proposed in this thesis.In the establishment phase of the fingerprint database,by adopting this algorithm,the RSSI values firstly are measured several times at each reference position and are preprocessed in order to build the fingerprint database,which is then divided by using the improved K-means clustering based on simulated annealing.In the online positioning phase,the clustering sub-classes are found by calculating the distance between the points to be measured and the center of each cluster.Then,the WKNN algorithm is used to estimate the position coordinates of the points to be measured.The algorithm reduces the matching range of positioning and removes the interference of the singular point to some extent,thus reducing the positioning error.Finally,the stability of the localization algorithm based on simulated annealing and K-means clustering is proved to be stronger than that K-means clustering,which will not be affected by the initial clustering center.Compared with the original WKNN algorithm,thisimproved algorithm could improve the positioning accuracy under the premise of stability of positioning.