基于三维Voronoi图划分的加权混合回归定位算法

Weighted mixed regression localization method based on three-dimensional Voronoi diagram division

随着无线通信技术和感知技术的发展,基于无线传感器网络的各种技术应运而生,这些技术被广泛应用在智慧农业、智慧交通、消防救援等领域。节点定位技术是无线传感器网络的基础技术之一,位置信息是感知数据的一部分,它决定了下一步要采取的具体措施。由于三维空间定位环境的复杂性,将平面上的定位方法应用在三维空间会有一定的局限性。针对以上问题,研究了基于三维空间Voronoi图的加权混合回归定位算法WMR-SKR。该定位算法分为离线训练和在线测试两个阶段。根据网络中的锚节点对定位空间进行三维Voronoi图划分,离线训练阶段将锚节点和Voronoi cell顶点的坐标组成的序列作为训练集进行训练。在线测试阶段通过训练好的定位模型对网络中未知节点的坐标进行预测。仿真实验结果表明,所提算法可有效降低三维空间中的节点定位误差,同时有效提高节点定位速度。

With the development of the wireless communication technology and sensing technology,various technologies based on wireless sensor networks are applied.These technologies are widely used in the fields of intelligent agriculture,intelligent transportation,fire rescue and so on.Node localization technology is one of the basic technologies of wireless sensor networks.Location information is a part of the sensing data,which determines the specific measures to be taken in the next step.Due to the complexity of the three-dimensional(3D)space localization environment,the application of the plane positioning method in 3D space will have some limitations.Aiming at above problems,the weighted hybrid regression location algorithm WMR-SKR based on a 3D Voronoi diagram was studied.The localization algorithm was divided into two stages:offline training and online testing.The 3D space was divided into Voronoi diagrams according to the anchor nodes in the network.In the offline training stage,the sequence composed of the coordinates of the anchor nodes and Voronoi cell vertices was used as the training set for training.In the online test stage,the coordinates of unknown nodes in the network were predicted through the trained localization model.Simulation results show that the WMR-SKR algorithm can effectively reduce the node localization error and improve the node localization speed in 3D space.