A novel algorithm for weighted average voting used in fault tolerant computing systems
2004; Elsevier BV; Volume: 28; Issue: 7 Linguagem: Inglês
10.1016/j.micpro.2004.02.006
ISSN1872-9436
Autores Tópico(s)Neural Networks and Applications
ResumoWeighted average voting has been widely used in control and computing systems. A novel algorithm for weighted average voting based on the use of a soft threshold is described. It has some benefits over the previously introduced weighted average voters, and is applicable for handling the output of an array of skewed sensors in safety related applications (such as weapons and transportation systems) as well as for arbitrating between the output of redundant clocks in highly available distributed computer systems. The performance of the novel voter is evaluated through a series of fault injection experiments and the results are compared with those of the well-known conventional Lorczak's weighted average voter. The experimental results showed that the novel voter gives better performance in terms of safety than the Lorczak's weighted average voter for small and large persistent errors.
Referência(s)