基于CSI相位差值矫正的室内定位算法

摘要

Aiming at the problems of phase error offset, high fingerprint noise and low accuracy of sample classification in most of the existing indoor passive fingerprint location methods based on Channel State Information (CSI) in complex scenes, an indoor fingerprint location algorithm based on Phase Difference (PD) value correction is proposed. In the offline period, the phase error and offset in the communication link are corrected by calculating the phase difference, and a robust fingerprint database is established. The BP neural network is used to train the fingerprint feature data to obtain the mapping relationship between fingerprint feature information and physical location. Online phase, this algorithm sets the phase difference filtering of CSI real time sampling value as the model, carries out positioning computing and finally gets the accurate location result. Experimental results show that compared with the existing fingerprint-based location method, this method has the advantages of significant denoising effect and high positioning accuracy.%针对现有基于信道状态信息的室内无源指纹定位方法在复杂场景中多数存在相位误差偏移、指纹噪声大、样本分类精度低的问题, 提出一种基于相位差值矫正的室内指纹定位算法.在离线阶段通过计算相位差值矫正通信链路中的相位误差和偏移, 建立鲁棒的指纹数据库, 使用BP神经网络对指纹特征数据进行训练, 得到指纹特征信息与物理位置的映射关系模型.在线阶段相位采样值经过差值矫正后作为模型的输入, 计算得到最终的精确定位结果.实验结果表明, 与现有基于指纹的定位方法相比, 该方法具有去噪效果显著、定位精度高的优点.