基于高斯混合模型的非视距定位算法简

针对无线传感器网络室内节点定位，在分析定位误差模型的基础上，结合高斯混合模型提出了一种无需先验知识的节点定位算法。利用高斯混合模型，对含有非视距误差的距离测量信息进行训练，以获得接近真实值的距离估计值。为取得高精度的定位效果，采用粒子群算法对期望最大化 (EM)算法进行优化。同时结合优选残差加权算法对所得的距离值进行定位估计，得出目标节点坐标估计值。仿真实验结果证实了算法的有效性。

GMM-based localization algorithm under NLOS conditions

Aiming at indoor node localizations of WSN, a node localization algorithm, where priori-knowledge is not necessary, was proposed. on basis of analyzing the error model, combined with Gaussian mixture model (GMM). By training the distance measurements containing NLOS errors, the more accurate range estimations can be obtained. For higher localization accuracy, the particle swarm optimization (PSO) was introduced to optimize the expectation-maximization (EM)algorithm. Finally, by using the residual weighting algorithm to estimate the distance, the estimation coordinates of target nodes can be determined. The proposed algorithm was proved to be effective through simulation experiments.