基于改进WKNN的室内定位算法

Indoor positioning algorithm based on improved WKNN

针对经典的NN算法,K近邻算法,加权K近邻算法中度量相似距离多为简单的欧氏距离,提出了将曼哈顿距离替代欧氏距离作为定位匹配的度量距离改进定位算法。其相比于之前的算法定位精度更高,响应速度更快,适合应用到室内定位相关研究当中。考虑到Wi-Fi信号易受噪声等外界不确定因素的影响以及移动终端接收信号强度指示与真实值存在偏差而导致定位精度不高的问题,引入卡尔曼滤波算法对高斯拟合后的接收信号强度指示数据进行误差修正,结合加权K近邻匹配算法进行定位有很好的效果。

Aiming at the classic NN algorithm,KNN algorithm,and WKNN algorithm,the measurement of similarity distance is mostly simple Euclidean distance.It is proposed to replace Euclidean distance with Manhattan distance.Distance is used as a measure of location matching to improve the location algorithm.Compared with the previous algorithm,it has higher positioning accuracy and faster response speed,which is suitable for indoor positioning related research.Considering that the Wi-Fi signal is easily affected by external uncertain factors such as noise and the deviation of RSSI of the mobile terminal from the true value,resulting in low positioning accuracy,this paper introduces the Kalman filter algorithm to the Gaussian fitting RSSI data for error correction,combined with WKNN matching algorithm for positioning has a good effect.