Research On Indoor Positioning Algorithm Based On Location Fingerprint Identification

Since the birth of the Global Positioning System GPS(Global Positioning System),outdoor positioning has brought great convenience to our production and life.However,GPS can only work in an empty outdoor environment.Faced with the complex and changeable indoor environment,it is difficult for GPS to accurately locate indoors.In recent years,with the continuous maturity of wireless communication technology and the continuous innovation of smart terminal equipment,wireless networks and smart phones have naturally integrated into our lives and become an essential part of our lives.Therefore,the widely distributed WiFi network and portable mobile phone terminals are the first choice for indoor positioning.The indoor positioning algorithm based on location fingerprint recognition with WiFi as the medium has become the research hotspot of indoor positioning technology at present with the characteristics of high positioning accuracy,low cost and simple implementation.On the basis of traditional indoor positioning algorithms,this paper makes research on indoor positioning algorithms based on location fingerprint recognition,improves and optimizes traditional positioning algorithms,and improves the positioning accuracy and speed of indoor positioning.The main research contents of this paper are as follows:(1)In the offline acquisition phase,Aiming at the phenomenon that the intensity of the signal collected at the reference point is unstable due to the complex and changeable indoor environment,a mixed filtering process of directional filtering + Gaussian filtering is performed on the collected signal.The neighborhood filtering algorithm was introduced to the phenomenon of individual abnormal noise points in the fingerprint database after filtering,and the influence of different signal strengths generated by different APs(Access Points)on the center point was introduced.An AP-based AP was improved.The assigned neighborhood filtering algorithm makes the data source in the processed fingerprint database more stable.(2)In view of the large amount of data in the fingerprint database after the establishment,which leads to a large amount of positioning and matching calculations,the idea of cluster analysis in data mining is introduced into the division of the fingerprint database,and an experimental analysis is performed to pass the fingerprint database through a binary k-means clustering algorithm divided into appropriate clusters.In view of the fuzzy situation of fingerprint inventory at the clustering boundary point after using the binary k-means clustering algorithm,a binary k-means clustering algorithm based on double similarity is proposed by combining the similarity between the physical coordinates of the reference point and the similarity between the signal strength.Through this algorithm,the fingerprint database can be clearly and completely divided into three clusters,which improves the positioning accuracy and shortens the determination bit time.(3)Experiments and simulations are performed on common indoor positioning algorithms.Based on the experimental results,the WKNN(Weighted K-Nearest Neighbor)positioning algorithm is finally determined as the main research algorithm in this paper,and the following optimizations are performed on the WKNN algorithm.First,at the stage where the signal strength value of the positioning point is received,each AP is given a certain weight to determine the size of the contribution to the positioning,and the positioning point changes the neighboring point at different K values.A WKNN algorithm based on AP weighted transformable K-values is proposed.An improved algorithm was proposed for clustering edge anchors,which effectively improved the accuracy of clustering edge anchors.Finally,through comparative experiments,the traditional WKNN positioning algorithm and the improved algorithm in this paper are compared and analyzed,which proves the effectiveness of the positioning optimization technology.