基于改进WKNN的位置指纹室内定位算法

位置指纹算法是目前解决室内定位问题的主要方法,指纹特征和匹配算法为影响算法精度的两大因素.针对室内复杂环境下Wi-Fi信号强度波动较大的现象,提出了基于方差的加权距离以改进WKNN算法.在离线特征提取阶段,选择了均值和方差两个特征值,既反映该采样点的RSS幅值,也反映该点RSS的波动情况;在线阶段,根据方差提出了加权距离进行相似度的计算,查找距离最近的K近邻点,并以实际环境下采集的数据验证了改进WKNN算法在RSS波动大的情况下对定位效果的改善,在综合考虑了AP组合的影响后,实现了误差均值为1.456m的定位效果.

An Indoor Location Fingerprint Algorithm Based on Improved WKNN

Location fingerprint is the main technique to solve the problem of indoor positioning, which is affected by the extraction of fingerprint feature and the matching algorithm.As the fluctuation of Wi Fi signal strength in complex indoor environment, a weighted distance based on variance to improve WKNN is proposed. On the offline feature extraction stage, the mean and variance of data set as a characteristic is selected, which can not only reflects the magnitude of the RSS of sampling point, but also the fluctuation.On line stage, a weighted distance based on variance is presented to calculate the similarity and find the nearest K neighbors points.In an actual Wi Fi environment, the improved WKNN algorithm is verified to improve the performance of the algorithm in the case of large fluctuation of RSS.Finally, after considering the impact of APs, the mean error of position is 1.456m.