Design And Implementation Of Image Aided RF Fingerprint Positioning System

With the rapid popularization and development of mobile intelligent terminal and wireless network,location-based services have received extensive attention.Due to the advantages of the indoor positioning technology based on WLAN that it only needs to measure the received signal strength,and does not need the additional wireless infrastructure,it has been widely used as an indoor positioning solution However,the indoor environment is more and more complex,the received signal intensity shows a high degree of complexity and variability,the accuracy of the positioning that only based on a single signal intensity will be seriously affected.In view of the problem,the paper adopts image aid to realize indoor positioning of radio frequency(RF)fingerprint.And the detailed analysis and research on improving the positioning accuracy and reducing the computational complexity have also been given out.Firstly,the paper proposes an improved K nearest neighbor method based on the Chebyshev distance.The average error of the algorithm is 2.10m,which can reach the precision of K nearest neighbor method.The calculation of Chebyshev distance only involves addition and subtraction,compared with K nearest neighbor algorithm,its computational complexity is lower.And the positioning accuracy of the algorithm is higher than the traditional MaD algorithm,CBD algorithm and VaS algorithm.Secondly,in consideration of the RF fingerprint indoor positioning technology based on WLAN,a kind of RF fingerprint indoor positioning system based on uniform image aided is proposed in the paper.The influence of the position of the image acquisition device for the positioning accuracy and the change of the workload are studied.The test results show that the average positioning error can reach 1.61m when the image acquisition device is located in front of the positioning area;the positioning error is 1.26m when the image acquisition device is located above the positioning area,the positioning accuracy of the latter is increased by 22%compared with the former.Compared with the RF fingerprinting,the average positioning errors of the two cases are improved about 0.55m and 0.85m respectively.The user only needs to match the reference fingerprint in the region by the uniform image aided,and the computational workload is reduced greatly.Finally,the universality that image acquisition devices are often located in front of the classroom is considered.In order to improve the accuracy of the uniform image aided RF fingerprint localization algorithm when the devices is located in front of the classroom,a uniform image aided technique based on non-uniform fingerprint database is proposed.The basic idea is establishing a non-uniform fingerprint database that the fingerprint density will increase with the increase of distance.Based on the uniform image aided radio frequency fingerprint localization algorithm,the uniform and non-uniform fingerprint database is used to match the user's location.The test results show that when the image acquisition device is located in the front of the classroom,the accuracy of the uniform image aided that adopts the non-uniform fingerprint database is up to 1.35m,which is improved by about 0.30m compared with using uniform fingerprint database.The workload will increase with the increasement of the number of reference fingerprints.The average positioning accuracy of the image aided RF fingerprint indoor positioning technology can reach 1.26m.Compared with RF fingerprint positioning based on a single signal intensity,the average positioning technology accuracy is increased by 40%,and the workload is reduced by 75%.