期刊名称：International Journal of Engineering and Computer Science
印刷版ISSN：2319-7242
出版年度：2013
卷号：2
期号：8
页码：2649-2658
出版社：IJECS
摘要：Several voting algorithms have been described to arbitrate the results of redundant modules in fault-tolerant systems. The inexactmajority and weighted average voters are widely used in control and safety-critical applications. Inexact majority voters require anapplication-specific ‟voter threshold‟ value to be specified, whereas weighted average voters are unable to produce a benignoutput when no agreement exists between the voter inputs. A major difficulty with inexact majority voters is the need to choose anappropriate threshold value, which has a direct impact on the voter performance. The problem of all documented weightedaverage voters is their inability to produce a benign output in cases of complete disagreement between the voter inputs. Both typesof voters are unable to cope with uncertainties associated with voter inputs originated from erroneous software, noisyenvironment, or noisy hardware modules.A voting scheme based on fuzzy set theory was introduced which softens the harsh behaviour of the inexact majority voter in theneighbourhood of the „voter threshold‟ and handles uncertainty and some multiple error cases in the region defined by the fuzzyinput variables. A set of fuzzy rules determines a single fuzzy agreeability value for each individual input which describes howwell it matches the other inputs. The voter is experimentally evaluated from the point of view safety and availability andcompared with the inexact majority voter in a Triple Modular Redundant structured framework. It is predicted that the fuzzy votercan be invaluable in at least two cases 1) as a substitute for the inexact majority voter in applications in which a small degradationin the safety performance of the system is acceptable at the cost of a large increase of its availability and a considerable decreaseof its benign outputs 2) when arbitrating between the responses of dynamic channels of control systems which may include someuncertainty.Automatic fuzzy parameter selection based dynamic fuzzy voter for safety critical systems with limited systemknowledge. Existing fuzzy voters for controlling safety critical systems and sensor fusion are surveyed and safety performance isempirically evaluated. The major limitation identified in the existing fuzzy voters is the static fuzzy parameter selection.Optimally selected static fuzzy parameters work only for a particular set of data with the known data ranges. Dynamic voter isdesigned in such a way that it can be plugged in and used in any safety critical system without having any knowledge regardingthe data produced and their ranges
