Software Trustworthiness Assessment Tool (STAT)

Warfighters are increasingly dependent on software systems that must be reliable and secure to perform their missions. There is a growing concern about system security because many current software systems are neither reliable nor secure. Software should ideally be secure from external agents and conditions accidentally or purposefully degrading system performance. In short the software must be trustworthy. A method is needed for the system administrator to assess the current security level of a system and gain insight into vulnerabilities, so risk can be actively managed rather than incorrectly assumed to be at a constant level. We are pleased to submit this proposal to design, develop, and evaluate a Software Trustworthiness Assessment Tool (STAT). We will focus on techniques for assessing the software trustworthiness of an integrated system during runtime by learning what behavior at a system level is “normal,” and by extension, trustworthy, and what is unusual and potentially untrustworthy. Current solutions are generally based on the detection and blocking of intrusions, computer viruses, and other malware. STAT instead detects problems that occur as the result of unanticipated interactions between trusted software applications that lead to untrustworthiness in the system as a whole.