Network intrusion detection system based on recursive feature addition and bigram technique

摘要: Abstract Network and Internet security is a critical universal issue. The increased rate of cyber terrorism has put national under risk. In addition, attacks have caused severe damages to different sectors (i.e., individuals, economy, enterprises, organizations governments). Intrusion Detection Systems (NIDS) are one the solutions against these attacks. However, NIDS always need improve their performance in terms increasing accuracy decreasing false alarms. Integrating feature selection with intrusion detection shown be successful approach since can help selecting most informative features from entire set features. Usually, for stealthy low profile (zero – day attacks), there few neatly concealed packets distributed over long period time mislead firewalls NIDS. Besides, many extracted those packets, which may make some machine learning-based methods suffer overfitting especially when data large numbers relatively small examples. this paper, we proposing based on method called Recursive Feature Addition (RFA) bigram technique. system been designed, implemented tested. We tested model ISCX 2012 set, well-known recent sets purposes. Furthermore, technique encode payload string into useful representation that used selection. propose new evaluation metric (combined) combines accuracy, alarm way helps comparing systems best among them. designed selection-based noticeable improvement using metrics.