What is the shape of your security policy?
Sven Türpe · 2009
This new paradigm defines security policies on cause-effect relations and models security mechanisms in analogy with pattern recognition classifiers. It augments the arsenal of formal computer security evaluation tools with new techniques. A causality model represents possible causes and effects; the causes include threats and the effects may be undesired. Target security policies derived from functional specifications select permitted causalities. Security mechanisms extract features from causes and effects and enforce mechanism-specific policies, approximating the target policy. Advantages of the classifier paradigm are the ability to generalize from incomplete information and examples, to measure classification error and mechanism performance, and to analyze mechanism ensembles and compositions. The classifier paradigm also offers a conception of problem complexity and suggests paying more attention to the impact of mechanisms rather than to their inner workings.