Research On Feature Extraction And Scene Identification In Wireless Channel Based On Graph Theory

The purpose of this study is to explore and identify the fingerprint features of the measured data and to verify its characteristics in different scenes.This paper includes five parts:the characteristics of the wireless channel;modeling and feature extraction;scene fingerprint identification;scene complex fingerprint identification;scenic area identification and matching.1.Based on graph theory,the features of wireless channel are introduced.With these research works,the features of wireless channel are defined.2.Based on the three scenes in the real channel measurement results,we employ digital signal processing and principal component analysis to extract the features of channel and modeling.The results show that the channel features of the model are in good agreement with the actual measured data.3.We introduce the neural network to focus on the identification of two scenarios to be measured.Based on the sample data of the off-line training and on-line identification matching,the scene to be tested is matched.The experimental results show that the identification model is effective and the learning adaptability is better.4.The identification of the composite scene in the channel is researched by clustering analysis.Comparing with the results of different sections,we come to a conclusion that the sections can be classified according to the number of regions based on fingerprint classification.The results show that this algorithm is effective in the channel division,identification and classification.5.Time series analysis and decision tree model are used to identify and match scenes in a region.The results show that the probability of misjudgment of the two channel samples provided is small.