全向指纹和Wi-Fi感知概率的WKNN定位方法

A method of WKNN positioning based on omnidirectional fingerprint and Wi-Fi sensing probability

针对室内环境下Wi-Fi信号强度衰减受人体影响较大且存在信号缺失现象的现状,该文提出一种基于全向指纹和Wi-Fi感知概率的加权K近邻定位方法,离线阶段构建顾及用户朝向和Wi-Fi感知概率的全向指纹库,在线阶段将全向指纹库中的感知概率用于定位过程。分别开展了基于方向识别、全向指纹和该文所提定位方法的实验,该文所提的方法在K为2时定位精度最高,平均定位误差为1.42m,标准差为1.04m,45%定位结果的精度优于1m,80%定位结果的精度优于2m。实验结果表明,该方法在定位精度和稳定性方面优于基于方向识别定位方法和基于全向指纹的定位方法。基于全向指纹和Wi-Fi感知概率的WKNN定位方法能够减少用户身体遮挡和信号缺失对定位的影响,可提高Wi-Fi指纹定位的精度。

In view of the attenuation of wireless fidelity(Wi-Fi)signal intensity is greatly influenced by human body and there is signal loss phenomenon in indoor environment.A method of weighted K nearest neighbors(WKNN)positioning based on omnidirectional fingerprint and Wi-Fi sensing probability was proposed in this paper.It constructed omnidirectional fingerprint database considering user’s orientation and Wi-Fi sensing probability in offline phase.Sensing probability stored in the omnidirectional fingerprint database was used for calculating coordinates in online phase.Methods based on direction recognition,omnidirectional fingerprint and the proposed method were conducted respectively.When K equals to 2,the proposed method achieved the best positioning performance with mean error of 1.42 meters and standard deviation of 1.04 meters;moreover,the accuracy of 45 percent of positioning results was better than 1 meter,the accuracy of 80 percent of positioning results was better than 2 meters.The experimental results showed that the proposed method greatly improved positioning accuracy and stability comparing to the method based on direction recognition and the method based on omnidirectional fingerprint.The WKNN positioning method based on omnidirectional fingerprint and Wi-Fi sensing probability could reduce impacts of user’s body sheltering and signal loss on positioning error and improve the accuracy of Wi-Fi fingerprint-based positioning.