基于动态射频指纹的调频定位方法

发布时间：2018-10-24 15:37

【摘要】：针对调频(FM)广播信号动态变化对定位性能产生较大影响的问题,提出了动态射频指纹的FM定位方法。该方法采用多元线性回归和神经网络方法,根据离线阶段各个参考点和部分较准点的调频广播信号强度的关系,使用在线阶段较准点的信号强度实时估计在线阶段参考点的信号强度。通过这两种方法建立具有自适应能力的动态射频地图,并使用贝叶斯估计方法对目标进行定位。实验结果表明,相对静态射频指纹模型:采用多元线性回归的射频指纹动态映射模型定位误差平均减少9.1%,采用神经网络的射频指纹动态映射模型定位误差平均减少36.3%,有效抑制了射频信号动态变化时变性对定位性能的影响。
[Abstract]:Aiming at the problem that the dynamic change of FM (FM) broadcast signal has great influence on the location performance, a FM location method for dynamic RF fingerprint is proposed. This method adopts multiple linear regression and neural network method, according to the relationship between the intensity of FM broadcast signal at each reference point and part of the punctuality in off-line stage. The signal intensity of the reference point in the online stage is estimated in real time by using the signal intensity of the online stage. The dynamic radio frequency map with adaptive capability is built by these two methods, and the Bayesian estimation method is used to locate the target. The experimental results show that, Relative static radiofrequency fingerprint model: the location error of RF fingerprint dynamic mapping model with multiple linear regression is reduced by 9.1on average, and the location error of radiofrequency fingerprint dynamic mapping model using neural network is reduced by 36.3cm, which is effectively suppressed. The influence of dynamic variation of RF signal on location performance.
【作者单位】：中国科学院计算技术研究所;湘潭大学信息工程学院;北京三星通信研究院;
【基金】：国家自然科学基金资助项目(61374214) 国家科技重大专项(2014ZX03006003-002) 国家863计划项目(2013AA12A201) 太原市—中关村合作专项(130104)
【分类号】：TN934.2

【相似文献】