The complexity of embedded systems has been growing in recent years. Competition in the international markets has increased the quality requirements of products which contain embedded systems. It has also required shorter development times and several customer specific versions of a product to be developed. As a result, the cost of testing has increased up to 50 % of the total manpower in a development project. Therefore, automation of the routine work included in testing of embedded systems and their software has been estimated to be significant in lowering the costs and increasing the quality of products. The focus of this thesis is on simulation based integration test automation of embedded communication software. The objective is to raise the automation level in host based integration testing and to make testing and debugging easier for a human tester. A set of basic principles for integration test automation of embedded communication software is developed and studied. The principles are based on the global controllability and observability of testing and for time simulation, which is used to reduce non-determinism appropriately to enable repeatable test sessions and to make testing easier for a human tester. The constructed integration testing platform enables test automation in test execution and test result analysis in a practical way. The platform is generic in the sense that it allows use of different implementation technologies for protocol simulations and test cases, and can thus form a backbone for the different kinds of protocol test systems. The integration testing platform has been exploited in several industrial development projects. Two different protocol software test systems are discussed to demonstrate the results. The evaluation indicates that the host environment is very applicable to testing of the logical correctness of embedded software and for integration test automation. The software under test and its operational environment can be controlled and observed in a flexible and efficient way. Test execution can be carried out in a repeatable manner and test results can be derived automatically. In addition, an unlimited tracing capability and automated regression testing are facilitated along with synchronous debugging of the distributed test system.
|Place of Publication||Espoo|
|Publication status||Published - 1997|
|MoE publication type||G3 Licentiate thesis|