Improved Software Testing for Open Architecture

Abstract

Applying traditional manual US Navy testing practices to OA systems will limit many benefits of OA, such as system scalability, rapid configuration changes, and effective component reuse. Pairing profile-driven automated software testing with test reduction techniques should enable these benefits and keep resource requirements at feasible levels. Test cases generated by operational profiles have been shown to be more effective than those developed by other methods, such as random or selective testing, and more resource-efficient than exhaustive approaches. This research effort increases the fidelity of the operational profile, creating an environment model referred to as a High-Fidelity Profile Model (HFPM) that can statistically describe individual software inputs. Samples from the HFPM's probability distributions can generate operationally realistic or overly-stressful test cases for software modules under test. This process can be automated and paired with output checking functions, enabling automated effective software testing, and potentially improving reliability. Such models would be ideal for US Navy Open Architecture (OA) software because of the defined interface standards. HFPMs can enable effective testing in software reuse applications and are ideal for testing multiple releases of maturing software. This research defines the HFPM, presents a methodology to develop, validate, and apply it.