摘要: |
通过分析数字调制信号功率谱及高阶谱特征,对高阶谱的求取方法作了改进,并在此
基础上提出新的特征参数,结合瞬时统计特征,采用支持向量机分类器,实现了AWGN信道下
数字通信信号的制式自动识别。仿真表明,所提取的特征参数具有较好的抗噪性能,对调制
参数的变化具有稳健性。考虑脉冲成形的影响,在信噪比大于12 dB时,单种信号最低
正确识别率大于985%,平均识别率达995%以上。 |
关键词: 调制方式识别 谱分析 特征参数提取 支持向量机 脉冲成形 |
DOI: |
|
基金项目:国家自然科学基金资助项目(60990320,60990323);国家高技
术研究发展计划(863计划)项目(2012AA012305) |
|
A novel identification method for digital modulations |
ZHOU Min,FENG Quan-yuan |
() |
Abstract: |
By analysing the characteristics of digital modulations′ power spectr
um and higher order spectral, the method to calculate the higher spectrum is im
proved. On this base, new feature parameters are proposed. Combined wit
h the instantaneous statistical feature parameter and support vector machine(SVM
) cla
ssifier, the automatic identification of digital communication signals in the AW
GN channel is achieved. Simulation results show that the extracted feature para
meters have good anti-noise performance as well as robustness to the chan
ge of modulation parameters.In consideration of the impact of pulse shaping, wh
en the SNR value is greater than 12 dB, the lowest correct identification ra
te of the single
-type signal is greater than 98.5%, and the average recognition rate is above 99
.5%. |
Key words: modulation identification spectral analysis feature extraction support vector ma
chine pulse shaping |