Runtime monitoring of security-critical systems

Abstract

Software products are becoming increasingly difficult to test due to their size. System implementations are becoming too large and difficult to reason about their correctness. However, properties the system should conform to can usually be formally expressed in a succinct and concise manner. We propose to use such properties, compiled directly into code to support runtime monitoring of security-critical programs. We envisage a system in which the property monitor will run together with the critical system to enable us to stop the program, fix the problem, or perform any other action, if any of the monitored properties are violated. Properties will be independent of the main code, and no modifications to the original code would be needed as these monitors would complement it. These properties will be thus used to identify potential anomalies introduced with monitors. The system will be tested on a number of security-critical systems to explore the different types of properties which can be expressed, and its effectiveness.