Simulation and Application of Voting Algorithms for Safety Critical Systems

The avionics system in safety-critical systems like aircraft, launch vehicles and missiles is responsible for several crucial functions. To satisfy the stringent safety requirements, redundancy in the avionics system configuration, in both hardware and software, is commonly employed. Voting algorithms are used to identify the module that is at fault and isolate it from the system. In this work, the voting requirement is considered for a quad redundant avionics system in the context of a launch vehicle. Different classes of voting algorithms, like averaging, smoothing and median-based voters were studied and evaluated through simulations. Typical data sets were fed as input samples and random errors were injected into the input samples to simulate actual error scenarios. The selection of output by the voting system was examined in each case. The novel history-based weighted averaging scheme was found to be the most suitable for our application, with the highest number of correct results.