*** 9661512 Tang This Small Business Innovation Research Phase I project investigates the feasibility of an approach that combines the operational profile, rare conditions, importance sampling, stress testing, and measurement-based dependability evaluation in the late testing phase or early operational phase, to quantify dependability for critical software that requires very high reliability (< 10-6) and availability (> 0.999999). Traditional software testing methods combined with reliability growth models cannot measure and assess dependability for software with such high requirements. The approach proposed for this research applies importance sampling, a statistical method to reduce sampling size while keeping estimates obtained from the sample unbiased at a high level of confidence, to the operational profile to guide testing critical operations or components of the software to accelerate the occurrence of rare conditions, then transforms the failure rates measured in the testing to those that occur in the normal operation by the likelihood ratio function of the importance sampling theory, and finally evaluates reliability and availability for the tested software system based on the transformed failure rates using measurement-based dependability modeling techniques. When the acceleration factor (likelihood ratio) is large (over 100), which is typical because the occurrence probability of rare conditions in the normal operation is much lower than in the importance sampling testing, it is possible to quantify a very high reliability or availability in a test of reasonable duration. It is anticipated that upon the completion of this research, a detailed methodology and a high level design of tools for implementing the proposed approach will be available. Real-time commercial production systems that use software to monitor, control, and manage safety-critical equipment and physical processes (e.g., aircraft, nuclear and chemical plants) and similar critical systems developed by government agencies (e.g., space shuttle and air traffic control system) require high reliability, availability, and safety. The approach and tools addressed in this research can be applied to the software in these systems to quantify dependability and to provide feedback to the design. The approach can also contribute significantly to enhancing the quality of the systems by applying its stress testing methods to the software. ***
