基于改进粒子滤波算法的无线传感器网络节点定位

Node localization in wireless sensor networks based on improved particle filter algorithm

针对无线传感器网络节点定位过程中的非视距传播误差问题,提出1种改进的状态检测粒子滤波算法。引入节点随机运动模型对节点运动状态进行预测。通过马尔科夫过程对在非视距(NLOS)/视距(LOS)混合环境下获得的测量值进行检测。利用p-范数对NLOS测量值进行筛选。结合校正后的节点间测量值和节点真实移动速度构建锚盒和采样盒。仿真结果表明,当NLOS/LOS混合模型分别满足均匀分布、高斯分布和指数分布时,该文算法均有较高的定位性能。

An improved state detection particle filter algorithm is proposed to solve the problem of non-line-of-sight error in the node localization of wireless sensor networks（WSNs）. Nodes＇ motion state is predicted by a random walk mobility model of the nodes. The measurements between nodes in the non-line-of-sight/line-of-sight（NLOS/LOS） mixed situation are identified by Markov process.The measurements including the NLOS error are selected according to the p-norm expression. An anchor box and a sampling box are built according to adjusted measurements between nodes and nodes＇ true mobile speeds. Simulation results indicate that,the proposed algorithm has high position accuracy when the NLOS/LOS hybrid model satisfies the uniform distribution,the Gaussian distribution and the exponential distribution respectively.