结合极大似然距离估计的MDS-MAP节点定位算法  

Localization Algorithm for Wireless Sensor Networks Based on MDS-MAP Integrated with Maximum Likelihood Estimating

作　　者：李津蓉[1,2] 王万良[1] 介婧[2] 姚信威[1] 

机构地区：[1]浙江工业大学计算机科学与技术学院,杭州310023 [2]浙江科技学院自动化与电气工程学院,杭州310023

出　　处：《传感技术学报》2016年第4期572-577,共6页Chinese Journal of Sensors and Actuators

基　　金：国家自然科学基金项目(61379123);国家自然科学基金项目(61402414);浙江省自然科学基金项目(LQ14F050002)

摘　　要：基于多维定标的定位算法通常利用节点间的最短路径长度代替欧式距离构建距离矩阵,当网络拓扑结构不规则时,会导致较大的定位误差。针对这一问题,提出了一种结合极大似然距离估计和多维定标的节点定位算法MDS-MAP(MLE)。算法将待测节点的一跳邻居节点信息作为极大似然方法的输入,利用与邻居节点的距离信息计算待测节点的相对坐标,然后根据已知锚节点的坐标,将所有节点的相对坐标映射为绝对坐标。实验结果表明,针对规则网络和不规则网络,MDS-MAP(MLE)算法均可取得较好的定位精度,且当网络连通度在一定范围内变化时,定位误差可保持在较低的稳定区间内。The multidimensional scaling（MDS）positioning algorithms of wireless sensor networks usually use thelength of shortest path in lieu of the Euclidean distance between two nodes to build the distance matrix,which mayresult in large positioning errors,especially when the network topology is irregular. To solve this problem,a new lo-calization algorithm MDS-MAP（MLE）was proposed in this work,in which the MDS-MAP method and MaximumLikelihood Estimating was combined.The information coming from the 1-hop neighbors are deemed as the input ofthe maximum likelihood method to estimate the relative coordinate of the tested node. And this process will iterateuntil all unknown nodes＇ relative coordinate are estimated. And then the global map can be transformed to an abso-lute map based on the absolute positions of anchors. Simulation results shows that the MDS-MAP（MLE）algorithmcan get high positioning precision,either for regular or irregular network. In addition,the positioning errors can stayrelatively constant at a low level when the network connectivity vary within a certain range.

关 键 词：无线传感网络 定位算法 多维定标 极大似然方法 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]