Behavior Signature for Fine-grained Traffic Identification

摘要: With the rapid development of Internet and a vigorous emergence new applications, traffic identification has becom e key issue for efficient network management. Although vario us methods have been proposed, there are still several limitations to achieving fine-grained application-level ident ification. In this paper, we propose signature model cal led behavior that utilizes inter-flow relation application traffic. The proposed is unique pattern appearing in first few packets plural flows when specific function cond ucted by an with combination various optional features. This contrast other existing signatur models usually focus on singular packet or flow feature extract ion identification. We proved feasibility applicability developing extraction algorithm conducting experime nts popular applications.