Abstract:  Received signal strength is one of ranging methods with low-power and low-cost, and used to estimate the location of nodes in wireless sensor network. The localization accuracy is affected by the time and space media, the fast algorithm is one of methods, which can improve the localization accuracy. The fast grid localization algorithm based on granular computing is proposed by forming the grid of localizing area. The localization problems are transformed into classification problems, and the granular computing classification algorithm is used to obtain the parameters related to the localization, which are used to estimate the locations of blind nodes. The experimental results show that grid localization algorithm based on granular computing can reduce the localization error and time consuming compared with localization by support vector machines.

Key words: Wireless Sensor Network, localization, granular computing

摘要： 接收信号强度作为一种低功率廉价的测距方式而用于估计无线传感器网络的节点位置，但定位精度会受到时空传播介质的影响，构造快速算法是解决该问题的主要方法之一。对定位区域网格化，提出了基于粒计算的快速网格化定位算法。将定位问题转化为分类问题，利用粒计算分类算法，得到定位参数，估计未知节点的位置。实验结果表明与支持向量机定位相比粒计算网格化定位算法降低了定位误差和时间。

关键词: 无线传感器网络, 定位, 粒计算