【摘要】 本文提出了一种基于超球粒计算的节点定位算法,超球粒计算一次扫描训练集即能完成训练过程,具有快速的训练特点,适合无线传感器网络的节点定位问题。首先,依据无线传感器节点的位置的相关参数信息,构建出多输入、多输出问题的训练集,其次,通过网格化方法划分定位区域,把原始训练集转化为多输入、单输出的分类问题的训练集,最后由超球粒计算算法完成扫描训练,从而得到有关参数,并估算未知节点的位置。实验仿真结果表明提出的算法取得较好的定位精度。更多还原

【Abstract】 Granular computing is a fast training algorithm and is suitable to the localization problem of wireless sensor network because it scans the training set by a single past. A localization algorithm is proposed based on hypersphere granular computing. Firstly,the training set with multi inputs and multi outputs is formed by the received signal strength between the anchor nodes and the reference nodes and the location of reference nodes. Secondly,the training set with multi inputs and multi outputs is transformed into the training set with multi inputs and single outputs by the grid method of localization area. Finally,the parameters of localization are achieved to estimate the location of unknown nodes. Experimental results showed that the proposed algorithm achieved better positioning accuracy.更多还原