基于近邻传播算法的动态自适应室内指纹定位算法

发布时间：2018-03-09 11:45

本文选题：近邻传播算法　切入点：方差滤波　出处：《计算机应用研究》2017年10期 　论文类型：期刊论文

【摘要】：目前传统的室内指纹定位算法中存在以下几个问题:首先在构建指纹库时采用平均值的方式容易受到噪声点影响而降低定位精度;其次使用欧氏距离衡量待定位点与指纹点之间的距离可能引入信号强度距离较近、物理距离较远的参考点参与估计待定位点的位置,从而增大定位误差,以及当参考点数量较大时,由于K近邻算法的计算量较大,造成定位过程耗时较大,能源耗费较多的情况;最后由于K近邻算法无法根据实际情况确定参与定位的参考点个数而限制了定位系统的精确性和拓展性。针对上述问题,设计了一种基于近邻传播算法的动态自适应室内指纹定位算法。该算法在离线阶段对在每一个参考点采集的信号强度值使用方差滤波算法去除噪声值,然后利用加入了参考点物理信息的近邻传播算法对参考点进行聚类处理。在在线阶段,通过进行粗略定位和精确定位动态地估计待定位点的物理位置。经过实验证明,所提出的新算法较对比算法有较高的精确度和稳定度。
[Abstract]:At present, there are several problems in the traditional indoor fingerprint localization algorithm: firstly, the average value method is easy to be affected by noise points and the location accuracy is reduced when constructing fingerprint database; Secondly, using Euclidean distance to measure the distance between the undetermined site and the fingerprint point may lead to the closer distance of signal intensity, and the reference point which is far away from the physical distance may participate in the estimation of the location of the undetermined site, thus increasing the positioning error. And when the number of reference points is large, because of the large amount of computation of K-nearest neighbor algorithm, the localization process is time-consuming and the energy consumption is more. Finally, the K-nearest neighbor algorithm can not determine the number of reference points involved in the location according to the actual situation, which limits the accuracy and expansibility of the positioning system. A dynamic adaptive indoor fingerprint location algorithm based on nearest neighbor propagation algorithm is designed, which uses variance filtering algorithm to remove the noise value of signals collected at each reference point in off-line phase. Then the reference points are clustered by using the nearest neighbor propagation algorithm with physical information of reference points. In the online stage, the physical location of the undetermined sites is dynamically estimated by rough location and accurate location. The proposed algorithm has higher accuracy and stability than the contrast algorithm.
【作者单位】：信息工程大学理学院;
【分类号】：TN92

【相似文献】