Applying parallelism to a bisimulation algorithm to improve efficiency in software testing of time-critical systems

Abstract

Time-Critical Systems (TCS) play a crucial role in environments where strict timing constraints are essential to ensure reliability and correctness. Mutation Testing (MT) is considered a valuable strategy for quality assurance of TCS, but it suffers from the equivalent mutant problem, which is known to increase computational cost and reduce confidence in MT. To address this problem, a strong timed bisimulation equivalence (STBE) algorithm can be used when TCS are modeled as Timed Automata (TA). STBE is computationally expensive and can benefit from parallelism. We survey available STBE implementations, identify opportunities to apply parallelism, build an extension that takes advantage of them, and test its effects. The resulting solution is a Java program that receives multiple TAs expressed in UPPAAL format and determines which TAs are equivalent using an STBE implementation such as TimBrCheck or MUTES and process-based parallelism. Compared to existing solutions, our tests show that our proposal is more efficient, reducing the runtimes of STBE by more than half. This could improve the reach, reliability, and effectiveness of MT for TCS.