《计算机应用研究》|Application Research of Computers

井下WLAN位置指纹样本自相关滤波降噪方法研究

Research on method of autocorrelation filtering and noise reduction for radio fingerprint samples of underground WLAN

免费全文下载 (已被下载 次)  
获取PDF全文
作者 宋明智,钱建生
机构 中国矿业大学 信息与控制工程学院,江苏 徐州 221000
统计 摘要被查看 次,已被下载
文章编号 1001-3695(2021)01-036-0179-05
DOI 10.19734/j.issn.1001-3695.2019.12.0636
摘要 为了解决井下WLAN位置指纹库中噪声样本对井下人员定位精度的影响,提出了用于消减位置指纹数据库中噪声样本的基于采样间隔<i>τ</i>的自相关滤波算法。研究表明,虽然在同一个参考点处使用同一个采集设备在不同时刻采集到的接收信号强度(received signal strength,RSS)序列都不完全一样,但噪声样本与其他样本的RSS序列相比有着较为显著的差异值。利用这一特点,基于采样间隔<i>τ</i>的自相关滤波算法使用样本均值作为两个样本间波动差值的平衡参照,使得噪声样本的异常特性被放大,进而使可能存在的噪声样本更精确地被滤除。滤除噪声样本的位置指纹库能够更好地表征各参考点处的RSS分布。实验结果表明,以90%置信概率为参照标准,使用自相关滤波后的位置指纹样本分别进行静态和动态人员定位的定位误差为3 m和3.5 m,比使用原始位置指纹样本的定位误差分别减小了1 m和0.5 m。
关键词 井下人员定位; 位置指纹; 自相关滤波; 样本波动差值
基金项目
本文URL http://www.arocmag.com/article/01-2021-01-036.html
英文标题 Research on method of autocorrelation filtering and noise reduction for radio fingerprint samples of underground WLAN
作者英文名 Song Mingzhi, Qian Jiansheng
机构英文名 School of Information & Control Engineering,China University of Mining & Technology,Xuzhou Jiangsu 221000,China
英文摘要 In order to solve the influence of noise samples in the underground WLAN radio fingerprint database on the personnel positioning accuracy, this paper proposed a new autocorrelation filtering algorithm based on sampling interval <i>τ</i>. The research shows that although the received signal strength(RSS) sequences collected by the same mobile terminal are different at the same reference point and different time, there are significant differences of RSS between noise samples and other samples. Using this feature, this filtering algorithm based on sampling interval <i>τ</i> was used the mean of the samples as the balance reference of the fluctuation difference between two samples, which enlarged the anomalous characteristics of the noise samples, so that the possible noise samples could be filtered more accurately. The radio fingerprint database which filtered out the noise samples could better represent the RSS distribution at each reference point. With 90% confidence probability, the experimental results show that the location errors of static and dynamic personnel positioning using radio fingerprint samples filtered by autocorrelation filtering algorithm are 3 m and 3.5 m respectively, which are 1 m and 0.5 m less than those using original radio fingerprint samples.
英文关键词 underground personnel positioning; radio fingerprint; autocorrelation filtering; sample fluctuation difference
参考文献 查看稿件参考文献
 
收稿日期 2019/12/9
修回日期 2020/1/27
页码 179-183,189
中图分类号 TP391.9
文献标志码 A