中南大学学报（自然科学版）2012，Vol.43Issue(3)：1020-1026,7.
基于支持向量机免疫集成预测的电信网络性能监控
Telecom networks performance monitoring based on artificial immune support vector regression
摘要
Abstract
To continuously monitor telecom network performance, an improved baseline generating method based on support vector machine (SVM) was proposed. An artificial immune support vector regression algorithm (AI-SVR) was presented, which optimized SVM parameters, kernel radius, embedding dimension and sample size by artificial immune network. The same point time series was built according to the cycle of telecom network performance and the baseline for telecom networks performance monitoring was defined as the predicted confidence interval. Taking the CPU load of a certain Softswitch server for analysis, the results show that AI-SVR can obtained a better regression mode than SVR with experienced parameters, the sum of error squares decreases by 55.4%, and the same point time series can overcome the problem that the output is sensitive to the abnormal input when using continuous series, and the monitoring method can find a few continuous anomalies.关键词
电信网络/性能监控/支持向量机/人工免疫系统/时间序列预测Key words
telecom network/ performance monitoring/ support vector machine/ artificial immune system/ time series prediction分类
信息技术与安全科学引用本文复制引用
郭建华,杨海东..基于支持向量机免疫集成预测的电信网络性能监控[J].中南大学学报（自然科学版）,2012,43(3):1020-1026,7.基金项目
国家自然科学基金资助项目(60973132) （60973132）
广东高校优秀青年创新人才培育项目(LYM09097) （LYM09097）
