一种改进的组合定权的指纹定位算法

室内场景复杂、WiFi信号不稳定等因素，造成基于信号空间K最近邻法的WiFi指纹定位算法匹配的邻近点会出现偏差，使用偏差较大的点计算待定点位置会直接影响定位结果。本文提出了一种改进的组合定权的指纹定位算法，对K个邻近点的几何结构进行分析，剔除其中偏离邻近点几何中心较远的点后，同时分析匹配邻近点中心同待定点几何位置存在理论上的关联，利用选择后的邻近点与其中心点的几何距离、待定点与指纹点欧氏距离组合定权，加权求取坐标。与KNN、WKNN算法定位结果分别进行比较，表明该方法提高了定位准确性和精度。

An Improved Fingerprint Indoor Localization Algorithm Based on Combination Weight

Under the influence of complex indoor scene,unstable WiFi signal and other factors,there are some mismatches in neighbor points selecting based on the algorithm of K-nearest.For these mismatch errors,there will be direct impact on the results of localization. In this paper, we proposed an improved fingerprint indoor localization algorithm based on combination weight. Firstly, analyze the geometrical structures of K-nearest points,and then remove the points which have the longest distance between the center of K-nearest points and itself.Secondly,analyze the geometrical location of unknown point and the center of K-nearest points,and use the geometrical distance between neighbor points and the center of them, the Euclidean distance between neighbor points and the unknown point to determine the weight of algorithm collectively.Finally,obtain the weighted localization results.Compared with the localization results of KNN and WKNN algorithm,the improved algorithm gets higher localization accuracy and robustness.