摘要
针对通信调制识别常用的决策树识别节点过多和支持向量机识别处理多类繁杂的问题,将二者结合起来采用决策树SVM算法,另引入在函数优化、数据挖掘等方面具有突出优势的混沌粒子群算法来对决策树SVM中参数选择进行择优提取。并对2ASK,4ASK,2PSK,4PSK,2FSK,4FSK这些数字调制信号,在信噪比为5%,10%,15%的情况下,进行仿真测试,结果表明有良好的分类效果。
To solve the inadequacies of the decision tree and support vector machine, the decision tree SVM to identify the communication modulation signal is introduced. The algorithm of chaotic particle swarm which has outstanding performance in function optimization and data mining is used to get the excellent parameters of decision tree SVM. When signal to noise ratio equal 5%, 10% or 15% , there is a simulation test about digital modulated sig- nals of 2ASK, 4ASK, 2PSK, 4PSK, 2FSK, 4PSK. The result shows that the decision tree SVM based on chaotic particle swarm performs very well in the classification of communication modulation pattern recognition.
出处
《电视技术》
北大核心
2012年第23期126-129,共4页
Video Engineering
基金
国家自然科学基金项目(61174107)
关键词
决策树
支持向量机
粒子群
混沌
调制识别
decision tree
SVM
particle swarm optimization
chaotic
modulation recognition
