基于SVM分类的位置指纹定位

Location Fingerprints Based on SVM

研究了基于位置指纹定位的室内定位方式,在分析了现有的改进定位算法后,结合实验环境,提出了一种基于支持向量机的位置指纹定位方法。实验中,将实验环境分成20个子区域。离线阶段,采集每个区域接收到的位置信号强度数据,建立位置信号强度和所在区域的关系,作为SVM训练样本集,得到最优分类模型。在线阶段,将实时采集到的位置信号强度作为测试集,采用SVM分类模型对其进行预测,判断所属区域。实验结果表明,提出的基于SVM分类的定位方法对以1m为间隔划分的区域具有良好的定位效果。

In this paper,indoor positioning method based on location fingerprinting is researched.After analyzing the existing improved positioning algorithm,combined with the experimental environment,a location fingerprinting method based on support vector machine is proposed.In this experiment,the experimental areas are divided into 20 sub-regions.During the off-line phase,the position signal intensity received in each area is collected,and the relationship between the signal intensity and the area is established as the SVM training sample to obtain the optimal classification model.In the on line phase,the position signal intensity collected in real time is taken as test set,using the classification model to predict its real time area.