Authors:
Oleksiy Mazhelis
and
Seppo Puuronen
Affiliation:
University of Jyväskylä, Finland
Keyword(s):
Wireless and mobile security, user verification, combining classifiers.
Related
Ontology
Subjects/Areas/Topics:
Enterprise Information Systems
;
Software Agents and Internet Computing
;
Telecommunications
;
Wireless and Mobile Computing
;
Wireless and Mobile Technologies
;
Wireless Information Networks and Systems
Abstract:
Modern personal mobile devices, as mobile phones, smartphones, and communicators can be easily lost or stolen. Due to the functional abilities of these devices, their use by an unintended person may result in a severe security incident concerning private or corporate data and services. The means of user substitution detection are needed to be able to detect situations when a device is used by a non-legitimate user. In this paper, the problem of user substitution detection is considered as a one-class classification problem where the current user behavior is classified as the one of the legitimate user or of another person. Different behavioral characteristics are to be analyzed independently by dedicated one-class classifiers. In order to combine the classifications produced by these classifiers, a new combining rule is proposed. This rule is applied in a way that makes the outputs of dedicated classifiers independent on the dimensionality of underlying behavioral characteristics. As
a result, the overall classification accuracy may improve significantly as illustrated in the simulated experiments presented.
(More)