Research On Auxiliary Frequency Based On Sequential Pattern Mining

Spectrum management is an important way to coordinate and manage the orderly use of frequencies.Radio monitoring is the basis of spectrum management,and a large amount of spectrum data generated in the monitoring process can provide effective support for management.Therefore,the information mining and knowledge acquisition of spectrum data have received extensive attention,especially the spectrum prediction technology based on sequential pattern mining has become a research hotspot in recent years.In the existing research,frequency spectrum prediction methods based on frequent pattern mining are mostly used in cognitive radio,and the deep application level of pattern information is low.In this thesis,the auxiliary frequency research based on sequential pattern mining is proposed.Frequent pattern mining using frequency spectrum data can not only carry out auxiliary frequency prediction,but also skillfully use the generated frequent pattern tree to evaluate the electromagnetic environment,further improving the reliability of auxiliary frequency.For FM broadcasting(87-108MHz),aviation intercom(118-137MHz),wireless intercom(403-424MHz)and WIFI(2400-2484MHz)four different service frequency bands,a long time of digital scanning spectrum data collection.The frequent pattern mining method like APriori algorithm is used for pattern mining,the frequent pattern is used for electromagnetic environment evaluation and spectrum prediction,and the auxiliary frequency is recommended to use channels.The main research work includes:(1)A signal-to-noise separation method is proposed.The noise is estimated by using the wave distribution in the spectral data and the adaptive calculation of the discriminant value is realized by using the information criterion.This method separates signal data from noise data by comparing wave height and discriminant value one by one without being affected by scanning step.Experimental results show that the algorithm can effectively separate noise and signal,and is applicable to most service frequency bands.(2)An electromagnetic environment evaluation method is proposed.The study of frequent mode tree shows that the electromagnetic environment of the channel is better when there are fewer leaves in the frequent mode tree and the tree depth is deeper,or the sum of weights of all nodes in the frequent mode tree is larger.Experiments are carried out in different service frequency bands,and the electromagnetic status of the channel can be evaluated accurately by means of frequent mode tree.(3)An auxiliary frequency method based on sequential pattern mining is proposed.The channel quality assessment,the channel electromagnetic environment assessment and the idle state of time series mode prediction are used to provide the auxiliary frequency decision.The experimental results show that the channel idle time concentration and frequency relative rule recommended by the proposed method meet the practical requirements.Finally,spectrum prediction can be carried out,which is predicted as idle frequency.Experimental results show that the recommended channel idle time concentration,the existing signal frequency,can accurately predict idle time.