Optimal allocation of test resources for software reliabilitygrowth modeling in software development

摘要

A component-based software development approach has become a trendnin integrating modern software systems. To ensure the overallnreliability of an integrated software system, its software componentsnhave to meet certain reliability requirements, subject to some testingnschedule and resource constraints. Efficiency improvement of thensystem-testing can be formulated as a combinatorial optimization problemnwith known cost, reliability, effort and other attributes of the systemncomponents. This paper considers "software component testing resourcenallocation" for a system with single or multiple applications, each withna pre-specified reliability requirement. The relation between failurenrates of components and "cost to decrease this rate" is modeled bynvarious types of reliability-growth curves. Closed-form solutions to thenproblem for systems with one single application are developed, and thenn"how to solve the multiple application problem using nonlinearnprogramming techniques" are described. Also examined are theninteractions between the system components, and inter-component failurendependencies are included in the modeling formula. In addition tonregular systems, the technique is extended to address fault-tolerantnsystems. A procedure for a systematic approach to the testing resourcenallocation problem is developed, and its application in a case study ofna telecommunications software system is described. This procedure isnautomated in a reliability allocation tool for an easy specification ofnthe problem and an automatic application of the technique. Thisnmethodology gives the basic approach to optimization of testingnschedules, subject to reliability constraints. This adds "interestingnnew optimization opportunities in the software testing phase" to thenexisting optimization literature that is concerned with structuralnoptimization of the software architecture. Merging these two approachesnimproves the reliability planning accuracy in component-based softwarendevelopment