基于聚类的组合时间反转算法的CSI指纹室内定位研究

Research on CSI Fingerprint Indoor Positioning Basedon Clustering-Based Combined Time Reversal Algorithm

基于信道状态信息(Channel State Information,CSI)指纹算法的室内定位,每个参考点存储的数据量越大,定位越准确,但是定位的耗时也更大。本文,创新提出了使用时间反转算法能量值进行聚类后,得到精简的指纹库,并保持定位的性能。首先离线阶段,对采集到的CSI数据进行误差的消除,并通过对各参考点的时间反转能量值进行二次聚类得到有效、精简的指纹库。位置估计阶段,提取未知地点的CSI数据,进行误差处理后与指纹库中的CSI信息进行多链路组合时间反转能量值的计算,估计位置。实验证明,该算法,在指纹库精简的同时,保持了定位的准确性,定位准确率达到90%以上。

Based on the indoor positioning of channel state information(CSI)fingerprint algorithm,the larger the amount of data stored at each reference point,the more accurate the positioning,but the time-consuming positioning is also greater.In this paper,the innovation proposes to use the time-reversal algorithm energy value to cluster,and obtain a simplified fingerprint database,and maintain the positioning performance.First,the offline phase eliminates the error of the collected CSI data,and obtains an effective and streamlined fingerprint database by performing secondary clustering on the time-reversed energy values of each reference point.In the location estimation phase,the CSI data of the unknown location is extracted,and the error processing is performed to calculate the multi-link combined time reversal energy value with the CSI information in the fingerprint database to estimate the location.Experiments show that the algorithm maintains the accuracy of positioning while the fingerprint database is streamlined,and the positioning accuracy rate is over 90%.