Abstract: Location fingerprints based indoor positioning, which uses wireless AP Received Signal Strength（RSS）, has become a popular research topic during the last a few years. Least Squares Support Vector Machines（LS-SVMs） based fingerprint-positioning is proposed in this paper. First, fingerprint based indoor positioning by LS-SVMs is given. Then, the detailed LS-SVM training process of fingerprinting samples is described. It focuses on how to transfer the positioning problem to a multi-class classification problem, which is handled by One-Against-One（OAO） and One-Against-All（OAA） approach respectively. Simulation results show that the proposed method has higher accuracy（average of 92.00%） and lower computational cost compared with traditional Support Vector Machines（SVMs） and k-Nearest Neighbors（K-NNs）.

Key words: location fingerprints, Least Squares Support Vector Machine（LS-SVM）, indoor positioning

摘要： 基于无线接入点（Access Point，AP）接收信号强度（Received Signal Strength，RSS）的位置指纹室内定位技术近几年已经成为国内外位置感知研究的热点。提出了基于最小二乘支持向量机（Least Squares Support Vector Machines，LS-SVM）的位置指纹定位方法。给出了基于LS-SVM的指纹定位模型，描述了LS-SVM指纹样本训练的具体实现过程。重点在于将定位问题转化为一个多类别分类问题，并分别采用一对一（OAO）和一对多（OAA）方法将其转化为多个二值分类问题。仿真结果表明，LS-SVM较传统支持向量机（SVMs）、K近邻（k-Nearest Neighbors，K-NN）定位方法的分类准确率高且计算代价小，平均分类准确率达92.00%。

关键词: 位置指纹, 最小二乘支持向量, 室内定位