基于多站点、多时间注意力机制的电磁强度时空关联分析与可视化  

Spatio⁃temporal correlation analysis and visualization of electromagnetic intensity based on multi⁃site and multi⁃temporal attention mechanism

作　　者：徐樱笑 资文杰 宋洁琼 邵瑞喆 陈浩 Xu Yingxiao;Zi Wenjie;Song Jieqiong;Shao Ruizhe;Chen Hao(College of Electronic Science and Technology,National University of Defense Technology,Changsha,410073,China)

机构地区：[1]国防科技大学电子科学学院,长沙410073

出　　处：《南京大学学报（自然科学版）》2021年第5期838-846,共9页Journal of Nanjing University（Natural Science）

基　　金：国家自然科学基金(61806211,U19A2058);湖南省自然科学基金(2020JJ4103)。

摘　　要：从监测端挖掘多监测站点间电磁信息隐含的时空关联对掌握电磁频谱资源分布和使用状况、支撑电磁频谱资源精细化管理具有重要意义.目前的方法孤立地对各监测站点采集的海量数据进行预测分析,没有有效挖掘多个站点间的电磁信息变化规律,没有考虑人类活动规律的周期性,更难以利用站点间的时域、空域关联给出多站点的电磁强度信息联合预测,不能结合GIS系统给出直观、实时可视化的频谱态势地图.针对以上问题,将多个监测站点建模为图结构,提出一种基于多站点、多时间注意力机制的图神经网络MMGCN(Multi⁃site and Multi⁃temporal Attention⁃based Graph Convolutional Neural Network),挖掘多站点电磁强度间隐含的时空关联,提升联合预测的准确性;基于多站点联合预测数据,设计一种并行克里金插值算法,对其他未布设监测点处的电磁强度信息进行空间插值生成电磁频谱态势地图,为电磁频谱精细化管理提供直观的态势预判与决策依据.Mining the spatial⁃temporal correlation of electromagnetic information from multiple monitor sites is of great significance for mastering the distribution and usage of electromagnetic spectrum resources.In traditional approaches,the prediction of electromagnetic intensity information mainly uses forecasting methods to predict and analyze the massive monitoring data collected by sites in isolation without correlation analysis in the time⁃domain or the spatial⁃domain.The prediction ignores to mine the law among multiple stations to give the joint prediction of electromagnetic intensity information and consider the periodicity of human activity law.Therefore,it is difficult to provide a real⁃time spectrum situation map based on the forecast data and GIS for managing electromagnetic spectrum resources.To address the challenges mentioned above,we firstly model the monitoring information by multi⁃site as a graph and propose a graph neural network based on multi⁃site and multi⁃temporal attention mechanism(MMGCN).MMGCN mines the implied spatio⁃temporal correlation between multi⁃site to improve joint prediction accuracy.Secondly,a parallel Kriging interpolation algorithm is designed to spatially interpolate the electromagnetic intensity information at other undeployed monitoring sites to generate a successive spectrum map,which provides intuitive situation prediction and decision⁃making basis for further management of electromagnetic spectrum.

关 键 词：多监测站点 多时间注意力机制 图神经网络 电磁强度预测 电磁频谱态势地图 

分 类 号：TP301[自动化与计算机技术—计算机系统结构]