An architecture-based comparison of verification and statistical reliability assessment methods for embedded software systems

Embedded computer systems are used to automate critical real-time applications, including process control systems, manufacturing systems, transportation systems, etc. In order to certify these critical systems, it is necessary to be able to rigorously demonstrate that high quality has been achieved. One approach is to decompose the software into more manageable aspects that can be independently certified. The software architecture is then used to certify the overall system. We compare the efficacy of formal verification and statistical reliability assessment methods for certifying the system based on the certification of the independent aspects and the system architecture.