基于图论算法的无线信道特征提取与场景识别研究

发布时间：2018-04-21 22:15

本文选题：图论 + 匹配　；参考：《海南大学》2017年硕士论文

【摘要】：本研究旨在发掘和鉴别所测得的信道数据的指纹特点,并证实其在不同场景的所具有的特性。本研究包括五部分内容:无线信道特性;建模及特性提取;场景指纹辨识;场景复合指纹辨识;场景区域辨识与匹配。1.基于图论知识介绍无线信道的特征。基于这项研究工作,明确无线信道特征。2.基于三种场景的真实信道测量数据,采用数字信号处理和主成分分析法对信道的特性进行提取和建模。研究结果说明,该模型体现的信道特征与实际测量的数据有较好的吻合度。3.引入神经网络重点研究两个待测场景的识别。通过对样本数据的离线训练与在线识别匹配,使得待测场景都获得匹配。实验结果表明,该辨识模型有效,学习的自适应性较好。4.采用聚类分析研究无线信道复合场景的鉴别。对不同路段的分析结果进行对比,得到结论:路段可以依据指纹划分的区域数量进行分类。研究结果表明,算法对信道的区分、辨识和分类的方法是有效的。5.采用时间序列分析和决策树模型对某区域的场景识别与匹配研究。结果表明,所提供的两个信道样本误判概率小。
[Abstract]:The purpose of this study is to explore and identify the fingerprint characteristics of the measured channel data and to verify their characteristics in different scenes. This research includes five parts: wireless channel characteristics; modeling and feature extraction; scene fingerprint identification; scene complex fingerprint identification; scene region identification and matching. 1. This paper introduces the characteristics of wireless channel based on graph theory. Based on this work, the wireless channel features. 2. 2. Based on the real channel measurement data of three scenarios, digital signal processing (DSP) and principal component analysis (PCA) are used to extract and model the channel characteristics. The results show that the channel characteristics of the model are in good agreement with the measured data. Neural network is introduced to study the recognition of two scenes to be tested. By off-line training of sample data and online recognition matching, the scene to be tested can be matched. The experimental results show that the identification model is effective and the learning adaptability is good. 4. 4. Cluster analysis is used to study the discrimination of wireless channel composite scene. The analysis results of different sections are compared, and the conclusion is drawn: the road sections can be classified according to the number of areas divided by fingerprints. The results show that the algorithm is effective for channel differentiation, identification and classification. Time series analysis and decision tree model are used to study the scene recognition and matching in a certain region. The results show that the error probability of the two channel samples is small.
【学位授予单位】：海南大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：O157.5

【相似文献】