期刊名称：International Journal of Computer Science and Network Security
印刷版ISSN：1738-7906
出版年度：2018
卷号：18
期号：2
页码：160-167
出版社：International Journal of Computer Science and Network Security
摘要：One of the key security challenges facing Optical Burst Switching (OBS) network that may influence its resource utilization performance, is flooding Burst Header Packets (BHPs). This problem is usually caused by edge nodes transmitting harmful BHPs that unnecessarily hold network resources causing the network to slowdown or in some cases deny the service. One emerging technology that may reduce this problem is the use of automated classification systems. These systems are based on data mining and are able to automatically label misbehaving nodes that send malicious BHPs before these BHPs impact network resources. This learning technology will not only save time and effort but also increase the accuracy of classification processes. In this paper, we investigate the applicability of rule induction nodes on the hard problem of BHP classification within OBS networks. Specifically, we propose a data mining model that adopts rule induction as a learning strategy to build classifiers that reduce flooding attacks by detecting misbehaving edge nodes early on. Empirical analysis using a recently published dataset reveals that data mining approaches generate promising results with respect to error rate. More importantly, the rule induction approach can detect nodes that are potentially transmitting malicious BHPs more accurately than other approaches such as those that are statistical and probability based.
