Optimization And Research Of Indoor Positioning Based On WiFi Location Fingerprinting Algorithm

With the rapid development of the Internet of things, more and more attentions are paid to location service. Because of the complexity of the indoor location environment, outdoor positioning technology can’t meet the demand of orientation. The arrival of the age of Internet makes the rapid expansion in WiFi network. With the advantages of low cost, easy deployment and wide spreading, WiFi technology becomes to an indispensable way for surfing the Internet in daily life. Position fingerprinting algorithm is a typical range-free algorithm, which establishes position fingerprint database in the first. Each position of the area is corresponding to a ’fingerprint’. Thus, the position of unknown point can be located by searching the fingerprint database. Therefore, based on the further study of existing WiFi location fingerprinting algorithm, this paper respectively points out the deficiency of the offline database construction stage and online positioning stage, and puts forward the corresponding improvement plan. Main task of this paper as follows:1. Considering of the noise of RSSI position fingerprint database, Gaussian model is proposed for measuring the received signal strength of sample nodes and unknown nodes to achieve the purpose of Filtering the value of low probability. K means clustering is put forward for data pretreatment, which divides location fingerprint database into K clusters. Every cluster stores the signal fingerprints of which the shortest Euclidean distance between each other. Cluster center is unique in every cluster. The pretreatment method can not only avoid the fingerprint clutter redundancy when data in large amount, but also reduce the amount of calculation in positioning phase by comparing the signal strength value of the unknown node and the cluster center.2. To solve the problem of high cost in updating the fingerprint database in terms of time and effort, the theory of Compress Sensing and Focus Lagrange Interpolation algorithm are proposed in the offline phase. The process of fingerprint vector refactoring is transformed into the problem of minimum l0 norm optimization by Compress Sensing, and total variation was used to recovery the original fingerprint vector. Focus Lagrange Interpolation algorithm takes the advantage of spatial correlation of sample nodes, by which the fingerprint database can be rebuild through measuring a small amount of fingerprints. Finally a practical experiment in real indoor environment shows the performance of Compress Sensing and Focus Lagrange Interpolation algorithm.3. On the basis of further research on the online positioning algorithm of existing indoor location fingerprint algorithm, this paper introduces an improved indoor weighted fuzzy matching algorithm to improve the positioning accuracy. This method transforms the problem of solving high order coordinates into the problem of space membership degree. The fuzzy neartude of unknown nodes and sample nodes can be calculated, which can determine the coordinates of unknown points. Compared with the traditional online estimation algorithms, this method reduces reduce the number of matching fingerprint participating in positioning, and distributes weights which are determined by the size of RSSI signal vector.Then coordinates of unknown points can be calculated by weights. Experiments show that positioning calculation process of weighted fuzzy matching algorithm is easier and positioning accuracy is higher.4. Kalman filter algorithm is used to correct error for estimates which is obtained by position fingerprint algorithm, and provide the real-time correction for location information of target in the process of movement. The result of experiment shows that the position accuracy of moving target improves significantly after filtering.Finally, Conclusion that shows main results of the search in this paper and points out the contents of further study.