Probabilistic optimization and assessment of voting strategies for X-by-wire systems

Signal voting of redundant sensor values and communication channels is of central importance in today's X-by-wire systems. The required degree of sensor redundancy, the type of redundancy, and finally the voting strategy must be designed to meet the system's dependability requirements. These design decisions depend on an analysis of the probabilities and effects of all underlying fault scenarios. Given a probabilistic fault model and a communication model, the voting step can be formally stated as a maximum-likelihood estimation of the correct input signal. With an example of an X-by-wire system we show how GTEFT can be used to derive the failure probabilities of different fault scenarios for various systems architectures and different voting strategies. Thus the capability of GTEFT to support system development and system assessment is demonstrated.