基于两种算法的无线信道“指纹”特征识别  

Recognition of Wireless Channel "Fingerprint" Feature based on Two Algorithms

作　　者：王瑞星[1] 刘斌[1] 杜健鹏[1] 李明[2] WANG Rui-xing LIU Bin DU Jian-peng LI Ming(Graduate School of Chinese Academy of Engineering Physics, Beijing 100088, China Xi＇an Aeronautics Computing Technique Research Institute, AVIC, Xi＇an Shaanxi 710068, China)

机构地区：[1]中国工程物理研究院研究生院,北京100088 [2]中航工业西安航空计算技术研究所,陕西西安710068

出　　处：《通信技术》2016年第10期1271-1279,共9页Communications Technology

摘　　要：针对无线信道"指纹"特征建模,包括"指纹"特征参数的建立、匹配识别、连续特征参数的"区域划分"等问题,用无线信道参数的提取算法、BP神经元网络算法和建立的微元试探法对模型进行分析求解。数值实验结果表明,矩阵奇异值分解和ESPRIT这两种无线信道参数提取算法对不同场景提取的参数能够进行很好地区分,BP神经元网络算法也能够准确识别场景和样本模式,利用微元试探法对连续信道区域的划分也证明是足够精确的。The wireless channel ＂fingerprint＂ feature is modelled, which involves the issues like establishment of characteristic parameters of ＂fingerprint＂, matching and recognition, and ＂region-dividing＂ of continuous characteristic parameters. The model is solved with the extraction algorithm of wireless channel parameters, BP neural network algorithm, and infinitesimal heuristics proposed by the authors. The numerical results show that based on parameters extracted in different scences by these two wireless channel parameters extraction algorithms including singular-value decomposition of matrix and ESPRIT, the several scenarios could be well distinguished, and meanwhile BP neural network algorithm could also accurately identify the scenes and sample models. Experiment indicates that the infinitesimal heuristics could make enough accurate division of the continuous channel region.

关 键 词：无线信道 指纹 BP神经网络 微元试探法 特征识别 

分 类 号：TN929.5[电子电信—通信与信息系统]