Abstract: Concerning the poor localization precision of Grid-Scan algorithm, an improved Grid-Scan localization algorithm based on virtual anchor nodes is proposed in Wireless Sensor Network （WSN）. Three related works of the improved algorithm are mentioned as follows：Firstly, setting the threshold of localizability for unknown node. The unknown node, the number of neighbor anchor node is greater than the threshold of localizability, is located by Grid-Scan, and the located unknown nodes are upgraded to virtual anchor nodes. Secondly, the number of neighbor anchor node which is greater than the threshold of localizability used the method of maximum likelihood to complete the localization, and the located unknown nodes are upgraded to virtual anchor nodes as well. Finally, anchor nodes and virtual anchor nodes participate in the localization of the remaining unknown nodes. Simulation results show that the improved algorithm, in difference of anchor node density, communication radius, grid size and irregular propagation models, has better localization precision.

Key words: Wireless Sensor Network（WSN）, Grid-Scan algorithm, localization precision, virtual anchor nodes, threshold of localizability, method of maximum likelihood

摘要： 针对无线传感器网络中Grid-Scan算法定位精度较低的问题，提出了一种基于虚拟锚节点策略的Grid-Scan定位算法。具体做了三个方面的工作：对未知节点设置可定位阈值，邻居锚节点数大于可定位阈值的未知节点使用Grid-Scan算法进行定位，定位后的节点升级为虚拟锚节点；邻居锚节点数小于可定位阈值的未知节点利用极大似然法完成定位，定位后的节点升级为虚拟锚节点；锚节点及虚拟锚节点共同参与对剩余未知节点的定位。仿真结果表明，改进算法在不同锚节点密度、不同通信半径和不同栅格大小的网络中以及通过不规则传播模型后都具有较好的定位精度。

关键词: 无线传感器网络, Grid-Scan算法, 定位精度, 虚拟锚节点, 可定位阈值, 极大似然法