| 高斯函数定权的改进KNN室内定位方法 |
| |
| 引用本文： | 毕京学,甄杰,汪云甲,刘笑笑.高斯函数定权的改进KNN室内定位方法[J].测绘通报,2017,0(6):9. |
| |
| 作者姓名： | 毕京学 甄杰 汪云甲 刘笑笑 |
| |
| 作者单位： | 1. 中国矿业大学, 江苏 徐州 221116;2. 中国测绘科学研究院, 北京 100830 |
| |
| 基金项目： | 国家重点研发计划，国家863计划，江苏省普通高校学术学位研究生创新计划 |
| |
| 摘 要： | 室内某些区域无线访问接入点（AP）布设稀疏，以及信号指纹的时变特性等因素，均使得无线信号接收信号强度（RSSI）序列与射电地图（radio map）相应RSSI序列完全相同成为可能，计算得到信号空间的欧氏距离为0或非常小。利用欧氏距离定权的加权质心算法解算会出现错误，无法得到定位结果；取K个参考点坐标均值的KNN算法以1/K为权值，定位精度相对较低。本文提出了高斯函数定权的KNN定位算法，对K个最近邻欧氏距离进行了标准化处理，利用高斯函数分配权值，得到加权坐标值。与KNN和WKNN算法的定位结果相比，该方法提高了鲁棒性和定位精度。
|
| 关 键 词： | 信号接收强度 欧氏距离 高斯函数 定权 K最近邻 室内定位 |
| 收稿时间： | 2016-09-20 |
The Method of Enhanced Gaussian Function Weighted KNN Indoor Positioning |
| |
| BI Jingxue,ZHEN Jie,WANG Yunjia,LIU Xiaoxiao.The Method of Enhanced Gaussian Function Weighted KNN Indoor Positioning[J].Bulletin of Surveying and Mapping,2017,0(6):9. |
| |
| Authors: | BI Jingxue ZHEN Jie WANG Yunjia LIU Xiaoxiao |
| |
| Institution: | 1. China University of Mining and Technology, Xuzhou 221116, China;2. Chinese Academy of Surveying and Mapping, Beijing 100830, China |
| |
| Abstract: | Because of less deployed APs in some indoor areas and signal fingerprint time-varying characteristics, it is possible for the currently scanning RSSI vector to be similar to corresponding RSSI sequence which is stored with location in radio map.In these cases, the calculated Euclidean distance is usually 0 or very small.Error will occur when the Euclidean distance is used for weight value in weighted centroid algorithm, and no result will be obtained.And KNN algorithm, which supposes the value 1/K as weight, will get the average of coordinates of K reference points along with relative low positioning accuracy.Therefore, Gaussian weighted KNN (GWKNN) localization algorithm is proposed: standardization processes for K nearest Euclidean distances were made, then corresponding weights were distributed by Gaussian function, at last, the weighted positioning result is obtained.Compared with the positioning results of KNN and WKNN algorithm, this positioning method can get higher robustness and positioning accuracy. |
| |
| Keywords: | RSSI Euclidean distance Gaussian function assign weights KNN indoor positioning |
| 本文献已被 CNKI 万方数据 等数据库收录！ |
| 点击此处可从《测绘通报》浏览原始摘要信息 |
| 点击此处可从《测绘通报》下载免费的PDF全文 |
|
