Current available testing methodologies have not been widely or productively utilized in industry for several reasons, including such factors as deadline pressures, necessity of a large initial investment in software and hardware, impracticality of methodologies, or inability to work with embedded systems. This project addresses one testing issue which has been problematic for industry: embedded systems. The inherent multiprocessing in embedded systems makes them particularly difficult to test. For example, each instrument control and display may be implemented as its own thread of execution. Testing all possible interleavings quickly results in a combinatorial explosion. Code complexity directly contributes to these combinatorics. One approach for handling complexity is through abstraction. This research seeks to develop a testing strategy based on state pools and constraints. Contraints represent relationships that must be maintained between objects. By treating these relationships as atomic units, execution granularity can be increased, reducing the number of interleaved execution paths. With the cooperation of an industry partner, application to production systems is being explored.
