Decentralized Gradient-based Field Motion Estimation with a Wireless Sensor Network
Decentralized Gradient-based Field Motion Estimation with a Wireless Sensor Network
复制标题
DOI：
10.5220/0005639100130024
复制
发表时间：
2016-02
期刊：
影响因子：
--
通讯作者：
D. Fitzner;Monika Sester
中科院分区：
文献类型：
--
作者：
D. Fitzner;Monika Sester
时空场平流的信息是预测或插值算法的重要输入。例如降水插值或预测算法，以及对由洋流平流的动态海洋学特征演变的预测。在本文中，提出了一种由固定且同步的无线传感器网络（WSN）节点对时空场运动进行分散式估计的算法。该方法基于著名的基于梯度的光流法，并扩展到无线传感器网络和时空场的特性，例如样本的空间不规则性、对计算和通信的严格限制以及在采样周期内假定的运动恒定性。给出了算法的规范以及对其通信和计算复杂性的全面分析。通过对传感器网络和时空运动场的模拟说明了该算法的性能。
Information on the advection of a spatio-temporal field is an important input to forecasting or interpolation algorithms. Examples include algorithms for precipitation interpolation or forecasting or the prediction of the evolution of dynamic oceanographic features advected by ocean currents. In this paper, an algorithm for the decentralized estimation of motion of a spatio-temporal field by the nodes of a stationary and synchronized Wireless Sensor Network (WSN) is presented. The approach builds on the well-known gradient-based optical flow method, which is extended to the specifics of WSNs and spatio-temporal fields, such as spatial irregularity of the samples, the strong constraints on computation and communication and the assumed motion constancy over sampling periods. A specification of the algorithm and a thorough analytical analysis of its communicational and computational complexity is provided. The performance of the algorithm is illustrated by simulations of a sensor network and a spatio-temporal moving field.