FUTA JOURNAL OF RESEARCH IN SCIENCE

VOLUME 13 NUMBER 2 2017

Paper Details

  • Title :BAYESIAN APPROACH TO NETWORK INTRUSION DETECTION ON WEB SERVER LOG DATA
  • Author(s) : 1B. A. Onyekwelu, 2A. O. Adetunmbi
  • Abstract:

The astronomical growth in global interconnectivity is a highly welcomed development, due to its positive impact on schools, organizations, institutions and individuals, especially with respect to access to critical information. However, this growth has also brought about increase in unauthorized access to critical enterprise information, leading to unforeseen risks in information management. This paper first discussed the Bayesian approach to Intrusion Detection, and the Architectural Model of a Bayesian Rate Intrusion Detection System. Preprocessed data obtained from Web Server Log of a University Network was used. Label Extraction and Data Set Balancing were carried out on the preprocessed data. Training and testing datasets where obtained, and a Bayes classifier, based on the Bayesian Information Criterion, was implemented on a System running Windows Operating system, using C++ programming language. The class labels were compared from the results obtained. it is clear from the results that Bayesian approach takes care of the problem of false alarms, and therefore, it is a standard tool for Intrusion Detection. The tools provide a means of solving the problems of data and information security.
Keywords: Intrusion Detection, Discretization, Bayesian Information Criterion,