基于KNN-SVM算法的室内定位系统设计

Indoor positioning system based on KNN-SVM algorithm

以室内的用户定位需求为应用背景,提高定位精度为目标,针对室内中复杂的环境,基于最近邻法(KNN)和支持向量机(SVM),提出了新的室内定位算法.先采用KNN去除训练样本中的奇异点,再采用支持向量机进行定位.与KNN法、朴素贝叶斯法、SVM回归法等室内定位算法比较,结果表明该定位算法有效提高了定位精度和定位速度.进一步提出了基于Android平台的室内定位系统的设计方案,采用Java语言编程实现了该系统,并进行了系统测试.实验数据表明:该室内定位系统的平均误差为1.7m,最大误差为4.9m,该系统在满足速度要求的前提下,有效提高了室内定位精度.

Aiming at the complex indoor environment,new indoor localization algorithm was proposed based on nearest neighbor algorithm(KNN)and support vector machine(SVM).The singular points in the training samples were removed by KNN,and then the support vector machine was used to locate it.The results show that the algorithm can improve the positioning accuracy and speed of the algorithm by comparing with KNN,Naive Bayesian and SVM regression method.The design scheme of indoor positioning system based on Android platform was put forward,and the system was realized by Java language,and was tested.The experimental data show that average error of the indoor positioning system is 1.7mand the maximum error of it is 4.9m.The system can effectively improve the indoor positioning accuracy under the premise of meeting the speed requirement.