基于小波核支持向量机的均衡器研究  

Equalizer based on wavelet kernel support vector machine

作　　者：许悦雷[1,2,3] 李璟旭[3] 马时平[1] 宋恒[4] 

机构地区：[1]空军工程大学工程学院 [2]西北工业大学电子信息学院 [3]北京航空航天大学电子信息工程学院 [4]中国人民解放军95661部队

出　　处：《系统工程与电子技术》2009年第9期2223-2226,共4页Systems Engineering and Electronics

基　　金：国家高技术研究发展计划(863计划)(2007AA701206)资助课题

摘　　要：针对通信系统实时性处理要求,提出了一种快速收敛的支持向量机(support vector machine,SVM)均衡器。该方法以支持向量机为框架,利用判决反馈信号构造SVM的训练样本数据,采用小波核函数,并自适应调整核函数中的伸缩因子,得到具有跟踪能力的自适应SVM均衡器。通过仿真实验,并与采用高斯核函数的支持向量机均衡器进行比较,结果证明该方法提高了收敛速度。To study the problem of the real-time data processing in communication system, a support vector machine equalizer with convergence speediness capability is presented. Based on the technology of support vector machine, train samples are constructed with decision feedback signals, and a wavelet kernel function is adopted and the flex parameter is adaptively adjusted in this equalizer. Then an adaptive equalizer with tracing abilities is designed. This novel equalizer is simulated in software and is compared with the support vector machine equalizer with Gauss kernel function. The result shows the proposed equalizer has much better performance in convergence speed.

关 键 词：均衡器 支持向量机 核函数 小波变换 

分 类 号：TN911.5[电子电信—通信与信息系统]