基于改进加权最小二乘支持向量机的UWSN定位  

UWSN Location Based on Improved Weighted Least SquaresSupport Vector Machine

作　　者：蒋华[1] 蔡晨 王慧娇[1] 王鑫[1,2] JIANG Hua;CAI Chen;WANG Huijiao;WANG Xin(College of Computer Information and Security,Guilin University of Electronic Science and Technology,Guilin 541004,China;College of Ocean Engineering,Guilin University of Electronic Science and Technology,Guilin 541004,China)

机构地区：[1]桂林电子科技大学计算机信息与安全学院,广西桂林541004 [2]桂林电子科技大学海洋工程学院,广西桂林541004

出　　处：《计算机测量与控制》2021年第8期250-254,共5页Computer Measurement &Control

基　　金：2019广西自然科学基金面上项目(2019GXNSFAA245053);广西科技重大专项(AA19254016)。

摘　　要：针对水下无线传感器网络锚节点较少、迭代误差大导致的节点定位精度低的问题,提出一种基于改进加权最小二乘支持向量机的水下三维节点定位算法;该算法将水下三维空间分为若干立方体,以锚节点与网格交点的距离向量作为训练集进行训练;并利用改进的多类别模式识别方法进行分类,以未知节点到锚节点的距离向量作为测试集确定节点坐标;通过引入加权的思想和多类别模式识别方法增大机器学习算法的鲁棒性、降低分类次数,从而实现水下三维节点预测定位;仿真结果表明,该算法在锚节点较少、网络区域较大的水下仍能保持较高的定位精度与较好的鲁棒性。Aiming at the problem of low node positioning accuracy caused by fewer anchor nodes and large iteration errors in underwater wireless sensor networks,an underwater three-dimensional node positioning algorithm based on improved weighted least squares support vector machine is proposed.The algorithm divides the underwater three-dimensional space into several cubes,and uses the distance vector between the anchor node and the grid intersection as the training set for training.Use the classification method of multi-class pattern recognition for classification.The distance vector from the unknown node to the anchor node will be used as the test set to determine the node coordinates.By introducing weighted ideas and multi-category pattern recognition classification methods,the robustness of machine learning algorithms is increased and the number of classifications is reduced,so as to achieve underwater 3D node prediction and positioning.Simulation results show that the algorithm can maintain high positioning accuracy even in underwater areas with fewer anchor nodes and larger network areas.

关 键 词：支持向量机 多类别模式识别 水下无线传感器网络 定位算法 

分 类 号：TP212.9[自动化与计算机技术—检测技术与自动化装置]