Dependability Evaluation of Complex Embedded Systems
Abstract: © 2010 IEEE. Personal use of this material is permitted. However, permission to reprint/republish this material for advertising or promotional purposes or for creating new collective works for resale or redistribution to servers or lists, or to reuse any copyrighted component of this work in other works must first be obtained from the IEEE.
The evaluation of the dependability performance (RAMS) of complex embedded systems requires the development of new approaches. In software-intensive systems, the dependability structure of the functions depends on the software. The search of fault sequences must involve software and hardware. The proposed method contributes to the qualitative and quantitative safety analysis of systems and micro-systems. Dependability ensures justified confidence in the ability of a system to deliver the services expected. Its metrics are multiple and complementary: availability, reliability, safety security, confidentiality, integrity, maintainability. They aim to ensure that under all circumstances and at all times, the operational capacity of the system (important for the functions of control and diagnosis, essential in the case of Safety Instrumented Functions) are maintained. The performance of an automated system (highly constrained embedded system), by its autonomy and the commitment hazard, must be durable and proven in terms of efficiency and productivity, dependability and respect for environment (energy, pollution). It is therefore necessary to develop methods and tools for integrate design of the analogic and digital subsets(continuous and discrete), hardware and software. This concurrent engineering problem has led to the development of modeling and validation techniques before entering into a collaborative design process. The exhaustive formal methods of proofs allow to reduce the risk of residual systematic errors. They can apply to both the hardware, and software architectures. The software will ultimately be realized by implementation in a processor. The dysfunctional properties of the couple software-hardware can not be reduced to the sum of the two parties. They are dependent on how the processor resources are requested by the programs and the reliability of these resources. The only way from our point of view, to model this interaction is to consider only the common object between software and hardware, which is the flow of processed information (as in the case of the methods a! pplied to the circulation of the fluids in the industry of process). The observation of this flow is used to identify the necessary resources and the consequences of their failures on the information conveyed. The modeling of these interactions can be realized by finite state automata  (IFD) whose languages contain all the sequences of events leading to alteration of the information output of the function. The paper is structured as follows: section 1 gives an overview on the problem of reliability evaluation of software / hardware pair. The methodology (modeling and reduction technique model) is presented in section 2. Two cases of application (Safety Instrumented System, microprocessor) are treated as third party.
Keywords: RAMS 2010 Proceedings - Reliability