摘要
为解决有记忆非线性的连续相位调制(CPM)信号调制方式识别精度低的问题,该文提出一种基于记忆因子的CPM信号最大似然调制识别新方法。该方法定义具有时齐马尔科夫性的映射符号,通过计算其后验概率构造记忆因子,进一步结合CPM分解和EM算法,推导出时间可分离,信道参数可估计的CPM信号似然函数。该调制识别方法所需符号数目少,适用信噪比范围广,识别CPM信号种类多且精度高,对相位误差鲁棒性强。仿真结果证明,当符号数目为200,信噪比为0 d B,相位误差任意时,该方法对8种CPM信号的识别率可达95%以上。
In order to solve the problem of low recognition accuracy of Continuous Phase Modulation （CPM） which is non-linear and with memory, a new maximum likelihood modulation recognition approach using memory factor is proposed in this paper. The approach defines the mapping symbol which has the time-homogeneous Markov property and generates the memory factor by calculating the posterior probability of the mapping symbol. Then, combining with the CPM decomposition and the EM Mgorithm, the time separable and channel parameter estimable likelihood function is deduced for CPM signals. The proposed approach has characters of low required symbol number, wide range of applicable SNR, large variety of recognizable CPM signals, high recognition accuracy and strong robustness to phase error. Simulation results show that the recognition rate of 8 kinds of CPM signals can reach more than 95% when the symbol number is 200, SNR is 0 dB and the phase error is arbitrary.
出处
《电子与信息学报》
EI
CSCD
北大核心
2016年第10期2546-2552,共7页
Journal of Electronics & Information Technology
关键词
信号处理
调制识别
连续相位调制
最大似然
EM算法
Signal processing
Modulation recognition
Continuous Phase Modulation （CPM）
Maximum likelihood
EM algorithm
