Forensic analysis of logs: Modeling and verification

  • Authors:
  • Mohamed Saleh;Ali Reza Arasteh;Assaad Sakha;Mourad Debbabi

  • Affiliations:
  • Computer Security Laboratory, Concordia Institute for Information Systems Engineering, Concordia University, Montreal, Que., Canada;Computer Security Laboratory, Concordia Institute for Information Systems Engineering, Concordia University, Montreal, Que., Canada;Computer Security Laboratory, Concordia Institute for Information Systems Engineering, Concordia University, Montreal, Que., Canada;Computer Security Laboratory, Concordia Institute for Information Systems Engineering, Concordia University, Montreal, Que., Canada

  • Venue:
  • Knowledge-Based Systems
  • Year:
  • 2007

Quantified Score

Hi-index 0.01

Visualization

Abstract

Information stored in logs of a computer system is of crucial importance to gather forensic evidence of investigated actions or attacks against the system. Analysis of this information should be rigorous and credible, hence it lends itself to formal methods. We propose a model checking approach to the formalization of the forensic analysis of logs. The set of logs of a certain system is modeled as a tree whose labels are events extracted from the logs. In order to provide a structure to these events, we express each event as a term of a term algebra. The signature of the algebra is carefully chosen to include all relevant information necessary to conduct the analysis. Properties of the model are expressed as formulas of a logic having dynamic, linear, temporal, and modal characteristics. Moreover, we provide a tableau-based proof system for this logic upon which a model checking algorithm can be developed. In order to illustrate the proposed approach, the Windows auditing system is studied. The properties that we capture in our logic include invariant properties of a system, forensic hypotheses, and generic or specific attack signatures. Moreover, we discuss the admissibility of forensics hypotheses and the underlying verification issues.