Hardware and software fault tolerance using fail-silent virtual duplex systems

Safety-critical systems must detect and tolerate hardware and software faults. The multiple virtual duplex system, the new scheme we propose for application in distributed control systems, efficiently covers both objectives. It comprises design and systematic diversity, time redundancy and a minimal amount of nodes. As a building block we use the virtual duplex system, which executes diverse variants of the software sequentially on a single node. For large control systems we offer two protocol types: the communication overhead can be kept low by a simple protocol, or can be slightly increased to enable a pipeline, leading to a drastic reduction in the required time.