摘要
为了解决同信道多信号的调制识别问题,提出了一种基于广义自回归(GAR)建模的调制识别方法。该方法利用观测数据的GAR模型参数估计各个待识别信号的短时平均中心频率和短时平均带宽,把一个多信号的调制识别问题转化为多个单信号的调制识别,并利用信号的短时平均中心频率和短时平均带宽的统计量作为特征输入到分类器,完成各个信号的调制类型识别。计算机仿真结果表明,当待识别信号在频域没有重叠或者部分重叠时,该方法都是有效的。
A generalized autogressive (GAR) modeling based modulation recognition method was developed to recognize the modulation types of multiple co-channel signals. By estimating the short-term average central frequency and short-term average bandwidth of each modulated signal from the GAR model parameters of their observed data, the modulation recognition of the multiple co-channel signals is then converted to multiple recognition of single signals. Statistics for the central frequency and the bandwidth are used as features to classify the modulation type of each modulated signal. Simulations show that the method is valid for both non-overlapped and partial overlapped co-channel signals in the frequency domain.
出处
《清华大学学报(自然科学版)》
EI
CAS
CSCD
北大核心
2009年第10期1676-1680,共5页
Journal of Tsinghua University(Science and Technology)
基金
国家"十一五"规划项目
关键词
调制识别
多信号
广义自回归模型
短时平均中
心频率
短时平均带宽
modulation recognition
multiple signals
generalized autogressive (GAR) model
short-term average central frequency
short-term average bandwidth