The most promising approach for developing secure systems is the one which allows software developers to assess and compare the relative security of their programs based on their designs. Thereby, software metrics provide an easy approach for evaluating the security of certain object-oriented designs. They can also measure the impact on security that caused by modifications to existing programs. However, most studies in this area focus on a binary classification of data, either is classified or

more »

... unclassified. In fact, there are other models with other classifications of data, for instance, the common model used by Defense departments that classifies data into four security levels. However, these various classifications have received little attention in terms of measuring their effect. This paper introduces a model for measuring information flow of security-critical data within a certain object-oriented program with multilevel classification of its security-critical data. It defines a set of object-oriented security metrics which are capable of assessing the security of a given program's design from the point of view of potential information flow. These metrics can be used to compare the security of programs or assess the effect of program modifications on security. Specifically, this paper proposes a generic model that consists of several security metrics to measure the relative security of object-oriented designs with respect to design quality properties of accessibility, cohesion, coupling, and design size.