误报率和漏报率影响入侵检测系统检测结果的可信性.通过从理论上分析误报和漏报产生的原因,提出了多检测系统协同工作提高检测可信度的方法.多检测系统结果融合时采用推进贝叶斯分类方法,给每个检测模型不同的权值,将分类结果加权求和,选择值最大的作为最终分类.实验分析表明,该方法降低了系统的漏报率和误报率,提高了报警的可信度.
False positive rate and false negative rate affected the detection creditability of intrusion detection systems (IDS). A method of multi-IDS cooperation to improve detection creditability was presented after analyzing false negative rote and false positive rate of IDS. The result fusion based on boosting Bayesian classification algorithm, which put different weights on single IDS and sum the result, then choose the greatest one. The experiments show that the method can reduce the false positive rate and false negative rate, then improve the detection creditability.