SVM在通信信号处理中的研究与应用

统计学习理论（statistical learning theory,SLT）提出结构风险最小化（structure risk minimization,SRM）准则,克服了人工神经网络（aretificial neural network,ANN）的模型过拟合和局部极小值问题.支持向量机（support vector machine,SVM）作为其具体实现方式,具有良好的泛化能力、高维处理能力和非线性处理能力.在简要介绍SVM原理的基础上,总结了其在通信信号处理领域的应用,包括数字调制识别、多用户检测、信道均衡、信号重构与相关参数估计、到达角（angle of arrival,AOA）与定位估计及其它6个方面,并指出了存在的问题和值得进一步研究的方向.

Research and Application of SVM in Communication Signal Processing

Statistical learning theory (SLT) proposes structure risk minimization(SRM) principle to overcome the over fitting and local minimum problems always existing in Aretificial Neural Network (ANN). As its concrete and successful implementation, support vector machine (SVM) has better generalization, high dimension and nonlinear processing ability. At first, theory of SVM is presented briefly. Then its applications in communication signal processing are summarized, including digital modulation classification, multiuser detection, channel equalization, signal reconstruction and parameter estimation, angle of arrival(AOA) and localization, etc. Finally, existent issues and future challenges are presented.