Spatio-temporal Information Mining Of Electromagnetic Targets Based On Radio Monitoring

As an important part of electromagnetic spectrum management,radio monitoring plays a significance role in protecting spectrum resources,maintaining the air radio wave order and guaranteeing the normal operation of various radio services within its jurisdicton.In recent years,due to the fast development of communication technology and increase of radio services,the electromagnetic environment has become more complex,radio monitoring has become more important,but also faces many new challenges.In military field,electromagnetism has been as a fifth domain of warfare,after land,sea,air and space.Electronic warfare is in an extremely important strategic position in modern warfare.It is very important for information warfare to quickly detect and precisely locate electromagnetic targets in complex environment,and to obtain their activity rule and comprehensive information in time.Therefor,studies on radio monitoring have profound pratical significance.This thesis focuses on solving the problems of response lag,location constraints and data resource waste existing in radio monitoring system.Based on the existing experimental equipment,a scheme of collecting radio spectrum data and electromagnetic targets' s spatio-temporal data is developed.Then,in the view of the the existing problems and requirements,methods of spectrum anomaly detection,electromagnetic target locating,and spatiotemporal data processing and mining are designed and applied to experimental data respectively.Finally,a prototype system for electromagnetic target spatio-temporal information mining is developed.The main research work and accomplishments are as follows.(1)The historical model of spectrum monitoring data is established,and an anomaly detection method based on distance is adopted to realize the anomaly detection of spectrum usage.This thesis studies an existing real-time detection method of spectrum anomalies based on Mahalanobis distance.In view of the shortcomings of this method,based on the segment similarity in time domain of spectrum,the original method is improved by introducing the strategy of segment modeling,which reduces the requirement of historical data volume in modeling and enhances the feasibility of the method.The new method is applied to the monitoring data of four different frequency bands and performes well.(2)After analyzing the outdoor experiment results of signal source direction finding,geographic information is introduced into the AOA(a locating algorithm using angle of arrival information)to improve the accuracy of location in real environment where existing NLOS propagation.Compared with traditional AOA,this method achieves better positioning accuracy in the outdoor experiment.(3)According to characteristics of spatio-temporal data of electromagnetic targets in passive wireless location,density-based clustering is introduced into spatio-temporal data processing to realize spatio-temporal clustering analysis and spatio-temporal frequent pattern mining.(4)A prototype system for electromagnetic target spatio-temporal information mining is developed,which mainly intergrates the algorithms that mentioned above.The system is capable of visualization of radio monitoring data,spectrum anomaly detection,location calculation of electromagnetic targets,spatio-temporal data mining and so on.