低信噪比下的数字调制识别算法研究

发布时间：2018-11-10 18:53

【摘要】：通信信号调制方式的识别技术已经经过了多年的发展,很多学者已经将各种技术应用到该领域,并取得了丰硕的成果。现在非协作通信仍然是军事和民用领域重要的通信形式。无线通信环境的复杂性和通信技术的发展都给调制识别技术提出了新的要求,在较低信噪比下进行高效识别成为了本文的研究重点。调制识别主要的工作是对接收的信号进行预处理,并提取特征参数,设计分类器对信号的调制方式进行识别。本文的主要工作和创新性成果如下:1.对基于瞬时信息提取特征参数的识别算法进行改进。提取了三个新的特征参数,分别是零中心归一化瞬时幅度绝对值的傅里叶变换最大幅值max?、归一化瞬时频率大于20的数量N、归一化非弱信号段瞬时频率的傅里叶变换最大幅值maxF。仿真性能表明改进的算法较传统方法和基于小波变换的调制识别算法更好。2.对基于信号功率谱提取特征参数的识别算法进行研究,详细分析信号的理论功率谱及其估计方法,提取有效的特征参数,并结合聚类算法实现信号的识别,该算法在信噪比大于5dB时识别率达到99%。3.对基于信号的高阶累积量的识别算法进行研究。在计算高阶累积量时对信号进行滤波,对不同的信号采用不同的方法求解高阶累积量,并修正了其他文献对4FSK微分信号高阶累积量的错误结果,实现的算法有效识别率在不低于0dB时满足实际的目标要求。4.对BP神经网络进行研究,并用本文中的五个特征参数作为网络的输入,对网络进行训练,结果表明本文所提取的五个特征参数与神经网络结合对调制信号进行识别,在低信噪比下实现了较高的识别率。
[Abstract]:The recognition technology of communication signal modulation has been developed for many years. Many scholars have applied all kinds of technologies to this field and have achieved fruitful results. At present, non-cooperative communication is still an important form of communication in military and civilian fields. The complexity of wireless communication environment and the development of communication technology put forward new requirements for modulation recognition technology. The main work of modulation recognition is to preprocess the received signal, extract the characteristic parameters, and design a classifier to identify the modulation mode of the signal. The main work and innovative results of this paper are as follows: 1. The recognition algorithm based on instantaneous information extraction feature parameters is improved. Three new characteristic parameters are extracted, namely, the number N of max?, normalized instantaneous frequency greater than 20, which is the largest value of Fourier transform, which is the absolute value of zero center normalized instantaneous amplitude. Fourier transform maximum value maxF. for normalized non-weak signal segment instantaneous frequency Simulation results show that the improved algorithm is better than the traditional method and the modulation recognition algorithm based on wavelet transform. 2. The recognition algorithm based on the characteristic parameters of the signal power spectrum is studied, the theoretical power spectrum and its estimation method are analyzed in detail, the effective feature parameters are extracted, and the recognition of the signal is realized by combining the clustering algorithm. When the SNR is greater than 5dB, the recognition rate of the algorithm is 99. 3. The recognition algorithm of high order cumulant based on signal is studied. The signal is filtered in the calculation of high-order cumulants, and different methods are used to solve the higher-order cumulants for different signals, and the error results of other literatures on high-order cumulants of 4FSK differential signals are corrected. When the effective recognition rate of the algorithm is not lower than 0dB, it meets the actual target requirements. 4. 4. The BP neural network is studied, and the five characteristic parameters in this paper are used as the input of the network to train the network. The results show that the five characteristic parameters extracted in this paper are combined with the neural network to recognize the modulation signal. A high recognition rate is achieved at low SNR.
【学位授予单位】：西安电子科技大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TN911.3

【相似文献】