On Enhancing Adaptive Random Testing for AADL Model

As the development of the large-scale and complicated software, especially in embedded system, non-functional properties of system, such as timing, reliability, safety and security, have become more and more important on impacting and restricting the behaviors of software system. One of the emerging challenges is how to test these properties in the phase of model-based software design. This paper aims to solve two essential problems in model-based testing: i) how to test model dynamically, ii) how to improve the efficiency of model-based testing. An enhancing adaptive random testing is investigated to generate test cases for AADL model-based testing in order to guarantee the system architecture and computing trustworthy. This methodology makes up the deficiency of adaptive random testing in dealing with the non-numeric data. A case study is presented and illustrates that its efficiency is higher than traditional random testing.