Software quality analysis by combining multiple projects and learners
Software Quality Control
VOMES: a virtual organisation membership evaluation system
International Journal of Networking and Virtual Organisations
Issues for robust consensus building in p2p networks
OTM'06 Proceedings of the 2006 international conference on On the Move to Meaningful Internet Systems: AWeSOMe, CAMS, COMINF, IS, KSinBIT, MIOS-CIAO, MONET - Volume Part II
| Hi-index | 0.00 |
Voting is a well-known technique to combine thedecisions of peer experts. It is used in fault tolerantapplications to mask errors from one or more expertsusing N-Modular Redundancy (NMR) and N-versionProgramming. Voting strategies include: majority,weighted voting, plurality, instance runoff voting,threshold voting, and the more general weighted k-out-of-nsystems. Before selecting a voting schema for aparticular application, we have to understand the varioustradeoffs and parameters and how they impact thecorrectness, reliability, and confidence in the finaldecision made by a voting system.In this paper, we propose an enumerated simulationapproach to automate the behavior analysis of votingschemas with application to majority and plurality voting.We conduct synthetic studies using a simulator that wedevelop to analyze results from each expert, apply avoting mechanism, and analyze the voting results. Thesimulator builds a decision tree and uses a depth-firsttraversal algorithm to obtain the system reliability amongother factors. We define and study the followingbehaviors: 1) the probability of reaching a consensus,"Pc"; 2) reliability of the voting system, "R"; 3)certainly index, "T"; and 4) the confidence index, "C".The parameters controlling the analysis are the numberof participating experts, the number of possible outputsymbols that can be produced by an expert, theprobability distribution of each expert's output, and thevoting schema. The paper presents an enumeratedsimulation approach for analyzing voting systems whichcan be used when the use of theoretical models arechallenged by dependencies between experts oruncommon probability distributions of the expert'soutput.