Research on Risk Probability Estimating Using Fuzzy Clustering for Dynamic Security Assessment

摘要

Effective network security management requires assessment of inherently uncertain events and circumstances dynamically. This paper addresses the problems of risk probability assessment and presents an alternative approach for estimating probability of security risk associated with some interaction in ubiquitous computing. A risk probability assessment formula is proposed, and an estimating model adopting the Fuzzy C-Means clustering algorithm is presented. An experiment based on DARPA intrusion detection evaluation data is given to support the suggested approach and demonstrate the feasibility and suitability for use. The practices indicate that fuzzy clustering technique provides concepts and theoretical results that are valuable in formulating and solving problems in dynamic security assessment.