论文部分内容阅读
检测器生成算法是影响人工免疫系统性能的重要因素之一,在大数据环境下由于自体数量的庞大使得现有检测器生成算法无法在有限时间内构建出成熟检测器集.在前期使用MapReduce模型构建分布式检测器生成系统的基础上,分析影响算法效率的主要因素;设计了MapReverseReduce模型构建检测器反向生成算法;通过Reverse阶段反转Map阶段的检查结果并将非法检测器键值对发送给Reduce阶段进行成熟检测器筛选,提高海量自体时人工免疫系统生成检测器的效率;最后在Hadoop集群中分别使用MapReduce模型和MapReverseReduce模型实现检测器生成算法的原型系统,并使用CERT synthethic sendmail data数据集进行测试与分析,验证了使用MapReverseReduce模型生成检测器的时间开销只有使用M apReduce模型时的5.22%~19.07%,并在自体数量不断增加时保持算法时间开销的稳定.
Detector generation algorithm is one of the important factors that affect the performance of artificial immune system. In the big data environment, the existing detector generation algorithm can not build a mature detector set in a limited time because of the huge amount of autocorrelation.Using MapReduce model Based on the construction of a distributed detector generation system, the main factors that affect the efficiency of the algorithm are analyzed. The MapReverseReduce model is designed to construct the detector reverse generation algorithm. The Reverse phase is used to reverse the result of the Map phase and send the illegal detector key-value pairs The maturity detector of Reduce stage was screened to improve the efficiency of detector for mass production of artificial immune system. Finally, a prototype system of detector generation algorithm was implemented in Hadoop cluster using MapReduce model and MapReverseReduce model, respectively. CERT synthethic sendmail data The test and analysis show that the time cost of generating detectors using MapReverseReduce model is only 5.22% ~ 19.07% when using M apReduce model, and keeps the algorithm time cost stable when the number of selfs increases.