Coverage-Driven Test Automation for Highly-Configurable Railway Systems.

Railway track systems contain nested variability, consisting of various configurable elements interacting in a configurable environment. Current test strategies rely on integration tests that simulate realistic operational scenarios defined by experts. Due to the increasing size and high configurability of railway software, the sole reliance on such test strategies is no longer feasible as the effort to reach test coverage increases exponentially. As a result, developers have to rely on slow testing processes that increase the development times of railway systems. In this paper, we leverage variability engineering techniques to create a procedure for automated, coverage-driven unit testing of railway systems. We generate sets of test cases using static code analysis on 150% models of track elements and propose different ways for automated test case execution in simulated test environments. We implemented our concept in a tool and applied it to railway projects provided by an industry partner. Our evaluation shows that we can automatically derive 150% models and compute test sets achieving branch coverage in a few seconds. Additionally, we show that an automated execution and evaluation of our test cases is possible in about 10 to 30 minutes, which is a significant speedup compared to current test strategies.