Identification of regression function and distribution model for denial of service attack in Second Life online community using simple network management protocol

Abstract

The evolution of internet results in the emergence of online communities. Numerous communities exist today with millions of users. Security, privacy and availability are the top constraints to be focussed in online communities. Among the other security violations, denial of service (DoS) ranks first as it disrupts the availability of services. DoS attack is reported in a popular virtual world, Second Life which made the complete portal inaccessible to legitimate users. As TCP-SYN is the more prevalent attack strategy of DoS, our solution is aimed to provide efficient detection of DoS in the Second Life community. Detection and differentiation of attack traffic from the legitimate is achieved through simple network management protocol (SNMP) and machine learning algorithms. In spite of the 'n' number of solutions, an efficient outperforming strategy with accurate attack detection is the critical requirement. This paper focuses on the DoS attack detection, classification using SNMP MIB variables and linear regression model. Experimental observation proves the detection accuracy of the method under DoS.