Using Partial Ordered Numbers to Control Information Flows

Abstract

Information flow control models can be applied widely. This paper discusses only the models preventing information leakage during program execution. In the prevention, an information flow control model dynamically monitors statements that will cause information flows and ban statements that may cause leakage. We involved in the research of information flow control for years and identified that sensitive information may be leaked only when it is output. However, most existing models ignore information flows induced by output statements. We thus designed a new model that especially emphasizes the monitoring of output statements. We also designed the model as a precise and low runtime overhead one. Our experiments show that the model bans every non-secure information flow and substantially reduces runtime overhead when comparing with our previous work.