Decomposition technique for integrated dependability evaluation of hardware-software systems using stochastic activity networks

Dependability evaluation is useful and important in designing computer systems, while at the same time it is also difficult. The difficulty becomes significant when the model combines hardware, software and their interactions due to the difference in failure behaviour between hardware and software. An approach to simplify the integrated evaluation of dependability is proposed using a structured top-down technique based on decomposition of system function in several levels. The sub-functions are then modelled using stochastic activity networks in terms of hardware, software and interaction. This in turn decomposes the system into components. The complete model is composed of the subnets of the sub-functions.