A study of effectiveness in masquerade detection

Abstract

Masquerade attacks are attempts by unauthorized users to gain access to confidential data or greater access privileges, while pretending to be legitimate users. Detection of masquerade attacks is of great importance and is a nontrivial task of system security. While several approaches do exist for masquerade detection, the relative effectiveness of approaches still needs considerable improvement. While in the past certain cost formulations have been used to compute the overall performance of masquerade detection methods, but these formulations appeared to be biased. Hence we present a formulation to compute the effectiveness of masquerade detection and also present a highly effective approach to masquerade detection using Hidden Markov Models (HMM). Our experimentation is on the well-known Schonalu dataset (SEA). Experimentation shows our approach to be most effective in the set of known approaches. © 2006 IEEE.