Robust ACCESS: Verification, Test, and Diagnosis of Advanced Scan Infrastructures

复杂的集成系统包含越来越多的非功能基础设施硬件，这些硬件支持甚至促进启动、维护、调试、测试和诊断。同时，所谓的可重构扫描网络(RSN)已经标准化，并且越来越广泛。虽然最初这种类型的基础设施专用于离线操作，但该项目续订将调查它们在运行时的重用，以支持安全关键系统中的容错和健壮性。该项目的第一阶段提出了验证、测试和诊断RSN的创新方法。这些技术现在将成为在运行时透明地使用RSN的基础，并满足用于故障检测和故障纠正的安全标准设置的严格的时间要求。例如，在汽车领域，安全关键系统包含已经可以自我测试或至少使用错误检测机制的核心。由于自检过程应将对整个系统的额外工作负荷降至最低，因此需要由非功能基础设施在后台收集错误指示器并将其传输到某个故障管理子系统的方法。这个子系统可以作为专用硬件实现，也可以作为专用协处理器实现，也可以只由一些软件实现，它负责对某些故障指示做出即时反应。对于故障消息的进一步适当操作，通常需要关于受影响的核心的更详细的诊断信息。额外的信息不应该通过在剩余的功能路径上增加额外的工作负载来获得，相反，将研究使用已经存在的非功能基础设施在后台交换诊断信息的方法。非功能基础设施只有在基础设施本身具有容错能力的情况下才能用于在安全关键系统中实现容错。在第一个项目阶段开发的RSN离线测试将扩展为在给定时间限制下的定期自检。将增加自诊断功能，以便即使在出现故障的情况下也能继续使用基础设施。将研究用于合成稳健RSN的方法，该方法允许在故障情况下在线重新配置RSN，以保证所有关键部件的可访问性。

Complex integrated systems contain an increasing part of non-functional infrastructure hardware, which support or even facilitate bring-up, maintenance, debug, test and diagnosis. So-called reconfigurable scan networks (RSNs) have meanwhile been standardized and are more and more widespread. While originally this type of infrastructure was dedicated for offline operation, this project renewal will investigate their reuse at runtime to support fault tolerance and robustness in safety critical systems. The first phase of the project came up with innovative methods for verification, test, and diagnosis of RSNs. These techniques will now be the basis to employ RSNs transparently at runtime and satisfy the strict temporal requirements set by the safety standards for fault detection and fault correction.For example, in the automotive area safety critical systems contain cores which are already self-testable or at least use error detecting mechanisms. Since self-test procedures should put a minimum of additional workload to the entire system, methods are needed to collect error indicators by the non-functional infrastructure in the background and to transport them to some fault management subsystem. This subsystem may be implemented as dedicated hardware, as a special coprocessor or just by some software, and it is responsible for the immediate reaction on some fault indication. For further appropriate actions upon a fault message, usually more detailed diagnostic information about the affected core is required. The additional information should not be obtained by putting additional workload on the remaining functional paths, instead methods will be investigated to use the already present non-functional infrastructure for exchanging diagnostic information in the background.The non-functional infrastructure can only be used for implementing fault tolerance in safety critical systems, if the infrastructure itself is fault tolerant. The offline test of RSNs developed in the first project phase will be extended to a periodical self-test following given time limits. Self-diagnostic features will be added to allow the continuing use of the infrastructure even in the case of faults. Methods for synthesizing robust RSNs will be investigated which allow the online reconfiguration of the RSNs in the faulty case to guarantee the accessibility of all critical components.