Information Fusion Implementation Using Fuzzy Inference System, Term Weight and Information Theory in a Multimode Authentication System

Abstract

Fuzzy inference systems have been used in a number of systems to introduce intelligence behaviour. In this paper we attempt to address an area of security challenges in identity management. The Sugeno-Style fuzzy inference is envisaged in the implementation of information fusion in a multimode authentication system in an effort to provide a solution to identity theft and fraud. Triangular and Sigmoidally shaped membership functions are used in the fuzzification of the three inputs categories namely biometrics, pseudo metrics and device based credentials. Term weight from text mining and entropy from information theory are used to compose the identity attributes metrics. Three corpora are used to mine the identity attributes and generate the statistics required to develop the metrics values from the application forms and questionnaires.Keywordsidentity attributes metricsinformation fusionfuzzy logicauthenticationterm weightentropy