Boundary Adherence and Exploration in High Dimensions for Validation of Black-Box Systems

State space exploration is limited to a testing budget in which a finite amount of resources can be expended to attain a degree of confidence in the system. A potential approach to increase the effectiveness of this budget is to focus testing on the boundary between targeted and non-targeted behavior, such as the boundary between where a function under test throws an error and executes normally. In this paper, an approach is presented that allows for tree-based exploration algorithms to adhere to the boundary between targeted and non-targeted behavior. This algorithm provides boundary points that fall within a predictable and controllable margin of error. Additionally, this boundary adherence strategy applied to a Rapidly Exploring Random tree for two, three, and twenty dimensions.