基于特征量重要度LS-SVR的WSN定位方法

针对无线传感器网络(WSN)节点定位方法中采用粗测距技术时,节点间较大的测距误差导致定位准确度不足的问题，提出一种基于特征量重要度LS-SVR的定位方法L-IFSVR.该方法把未知节点到锚节点的距离作为特征量,依据特征量的重要度进行特征提取,通过对探测区域网格化采样得到训练样本集,使用最小二乘支持向量回归机(LS-SVR)学习得到定位模型,在定位阶段,将未知节点的特征向量输入定位模型, 利用LS-SVR良好的泛化能力,实现对未知节点的准确定位.通过对均匀分布和C形区域随机分布的100个节点进行定位实验,结果表明,定位方法L-IFSVR能有效地降低测距误差对定位准确度的影响,减小平均定位误差,其中,均匀分布情况下L-IFSVR方法的平均定位误差相比采用相同测距技术的DV-Hop方法减小7.5~14.0%;C形区域随机分布情况下,显著减小36.5~55.2%

Localization in WSN Based on Importance of Feature LS-SVR

Aiming at solving the problem of low localization accuracy contributed by the big ranging error, which is the result of the coarse range technique adopted in WSN localization method, a new localization method based on importance of feature LS-SVR (L-IFSVR) is proposed. This method takes the range between the unknown node and anchor node as feature, and makes feature extraction according to the importance of the feature, gains the training samples through making gridding on the detection region, which are studied by using LS-SVR so as to get localization model. In localization phase, the feature vector of the unknown node is entered to the localization model, and finally the accurate location of the unknown node is achieved by utilizing the good generalization capability of LS-SVR. A localization experiment has been made on 100 uniformly distributed nodes and 100 randomly distributed nodes in C sharp region, whose result shows L-IFSVR can effectively lower the influence of the ranging error on localization accuracy and reduce the average location errors. In the uniform distribution, L-IFSVR made 7.5~14.0% less location errors than DV-Hop; in random distribution in C sharp region, the former made significantly 36.5~55.2% less location errors than the latter.