An Improved Fuzzy Voting Scheme for Fault ... - Semantic Scholar
International Journal of Applied Evolutionary Computation, 6(2), 41-49, April-June 2015 41
An Improved Fuzzy Voting Scheme for Fault Tolerant Systems Ram Murti Rawat, Indian Institute of Technology (BHU), Varanasi, India Tarun Kumar Gupta, Jawaharlal Nehru University, New Delhi, India Mohammad Sajid, Jawaharlal Nehru University, New Delhi, India Shiv Prakash, Department of Chemical Engineering, Indian Institute of Technology, Delhi, India Dinesh Prasad Sahu, School of Computer and Systems Sciences, Jawaharlal Nehru University, New Delhi, India Sohan Kumar Yadav, Jawaharlal Nehru University, New Delhi, India Chanchal Kumar, Jawaharlal Nehru University, New Delhi, India
ABSTRACT Voting is a widely used fault masking techniques for safety-critical systems to enhance the overall reliability of the system. Researchers over the period have proposed numerous advanced techniques in order to improve on the drawback of the existing methods. In this paper a fuzzy voting scheme has been survey and a generalized improved fuzzy voting scheme has been proposed. A comparative study of these schemes has also been carried out. It is found that proposed model is better than existing models. Single objective, multi-objective objective and many objective will be applied in future. Keywords:
Availability, Fault Masking, Fault Tolerance, Fuzzy Voting, Majority Voting, Safety, Threshold
1. INTRODUCTION To make a system ultra-reliable, fault tolerant design techniques are adopted using redundancies in hardware, software, information, time or any combination of these with an intent to minimise failure rate to an acceptable level. A failure in the system is characterised by
deviation of its performance with respect to the specification. These are due to presence of latent faults which could not be eliminated during design and development process. Fault masking enables correct functioning of a system even in presence of faults. Inspired by most successful society consensus system Voting is a extensively applied fault masking procedures
DOI: 10.4018/IJAEC.2015040103 Copyright © 2015, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited.
42 International Journal of Applied Evolutionary Computation, 6(2), 41-49, April-June 2015
for critical systems since it is simple to implement and can be adopted for both hardware and software. The technique originally suggested by Von Neuman has drawn huge attention of researchers resulting in new improved methods for selecting correct and reliable output. The cost and complexity are the factors generally used to determine the number of redundant modules to be used for consensus. The utmost popular voting procedures for use are inexact majority and standard weighted average. Those widely in use these algorithms suffer inherent disadvantages. On other hand majority voting is found to be excellent in safety performance. The availability of the system would be compromised due to low probability of correct output while the other algorithm ensures better availability and poor safety performance. Many improvements have been suggested to overcome the limitations of the above mentioned voting techniques (LatifShabgahi and Hirst, 2005). Fuzzy logic aims at modelling the imprecise modes of reasoning emulating human approximation approach to make rational decision in an environment of uncertainty and imprecision. Fuzzy approach is an alternate design methodology enabling solutions to complex problems using probabilistic logic and probability theory. Latif-Shabgahi et al. has proposed a fuzzy voting scheme for fault tolerant system (2005). The proposed voter can be practicable substitute to the traditional inexact majority voting algorithm and the results have been studied and analysed using Fuzzy tools. In this work, we propose an improved fuzzy voter which can be employed by critical applications such as flight critical system which demands ultra-high safety performance and availability. The organization of the remaining paper is given as follows. Section 2 reviews related work in the area and discusses the fuzzy voter scheme in detail. A modified improved fuzzy voter is proposed in section 3 enumerating its advantages. The simulation results of the proposed scheme have been compared for safety performance and availability with respect to existing scheme suggested by Latif-Shabgahi (2005) in section 4. Finally section 5 summaries the outcome of the study.
2. REVIEW OF RELATED WORK One of the furthermost extensively used fault masking techniques in massive redundancy system is to deploy a voter to mitigate the effects of the fault. Since the output of the redundant modules are likely to differ within it. Therefore Researchers over the period have proposed numerous advanced techniques in order to improve on the drawback of the existing methods (Das and Bhattcharya, 2010; Manic and Frincke, 2001). Classical Majority voting uses results of redundant modules to arrive at correct output based on the threshold value of output generally implemented on m of n system (Parhami, 1992; Ross, 2008). Exact voting method falls short for dealing with noisy sensors output and rounding off errors due to use of floating point arithmetic. To deal with these limitations, many variants of inexact voting have been proposed in the literature to handle variety of situations to suite application specific voter characteristics (Krstic et al., 2005; Parhami, 1994). Weighted average voting method arrives at consensus by assigning weights to the modules depending on the closeness of their outputs. This method improves the availability however it has an adverse performance for modules with larger output differences (Parhami, 1994; Prakash and Vidyarthi, 2014). A standard inexact majority voting defines a limit on the deviation of module output with respect to the actual correct result. This is widely used for safety critical application. This technique ensures high safety performance with constrained on availability factor. An improved scheme where the participation score of modules is maintained over the operational period which gives the number of occasions the module contributed in the voting process. In an event of majority consensus, the output of module with highest score is selected from the group of majority consensus participating module. A voting approach using both weighted average and history based adaptive voting has also been reported to have a better availability as compared to the algorithms discussed (Ross, 2008). These voters generally use fixed thresholds values to arrive at the consensus between the redundant
Copyright © 2015, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited.
7 more pages are available in the full version of this document, which may be purchased using the "Add to Cart" button on the product's webpage: www.igi-global.com/article/an-improved-fuzzy-votingscheme-for-fault-tolerant-systems/131447?camid=4v1
This title is available in InfoSci-Journals, InfoSci-Journal Disciplines Computer Science, Security, and Information Technology. Recommend this product to your librarian: www.igi-global.com/e-resources/libraryrecommendation/?id=2
Related Content Reliability-Aware Proactive Energy Management in Hard Real-Time Systems: A Motivational Case Study Satyakiran Munaga and Francky Catthoor (2012). Technological Innovations in Adaptive and Dependable Systems: Advancing Models and Concepts (pp. 215-225).
www.igi-global.com/chapter/reliability-aware-proactive-energymanagement/63583?camid=4v1a Optimal Kernel and Wavelet Coefficients to Support Vector Regression Model and Wavelet Neural Network for Time Series Rainfall Prediction B. Kavitha Rani and A. Govardhan (2014). International Journal of Applied Evolutionary Computation (pp. 1-21).
www.igi-global.com/article/optimal-kernel-and-wavelet-coefficients-tosupport-vector-regression-model-and-wavelet-neural-network-for-timeseries-rainfall-prediction/126208?camid=4v1a Complexity, Information, and Robustness: The Role of Information “Barriers” in Boolean Networks Kurt A. Richardson (2008). Intelligent Complex Adaptive Systems (pp. 187-214).
www.igi-global.com/chapter/complexity-informationrobustness/24188?camid=4v1a
What Makes Reality: Ontological Classes and Rules Azamat Abdoullaev (2008). Reality, Universal Ontology and Knowledge Systems: Toward the Intelligent World (pp. 76-109).
www.igi-global.com/chapter/makes-reality-ontological-classesrules/28311?camid=4v1a
An Improved Fuzzy Voting Scheme for Fault Tolerant Systems Ram Murti Rawat, Indian Institute of Technology (BHU), Varanasi, India Tarun Kumar Gupta, Jawaharlal Nehru University, New Delhi, India Mohammad Sajid, Jawaharlal Nehru University, New Delhi, India Shiv Prakash, Department of Chemical Engineering, Indian Institute of Technology, Delhi, India Dinesh Prasad Sahu, School of Computer and Systems Sciences, Jawaharlal Nehru University, New Delhi, India Sohan Kumar Yadav, Jawaharlal Nehru University, New Delhi, India Chanchal Kumar, Jawaharlal Nehru University, New Delhi, India
ABSTRACT Voting is a widely used fault masking techniques for safety-critical systems to enhance the overall reliability of the system. Researchers over the period have proposed numerous advanced techniques in order to improve on the drawback of the existing methods. In this paper a fuzzy voting scheme has been survey and a generalized improved fuzzy voting scheme has been proposed. A comparative study of these schemes has also been carried out. It is found that proposed model is better than existing models. Single objective, multi-objective objective and many objective will be applied in future. Keywords:
Availability, Fault Masking, Fault Tolerance, Fuzzy Voting, Majority Voting, Safety, Threshold
1. INTRODUCTION To make a system ultra-reliable, fault tolerant design techniques are adopted using redundancies in hardware, software, information, time or any combination of these with an intent to minimise failure rate to an acceptable level. A failure in the system is characterised by
deviation of its performance with respect to the specification. These are due to presence of latent faults which could not be eliminated during design and development process. Fault masking enables correct functioning of a system even in presence of faults. Inspired by most successful society consensus system Voting is a extensively applied fault masking procedures
DOI: 10.4018/IJAEC.2015040103 Copyright © 2015, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited.
42 International Journal of Applied Evolutionary Computation, 6(2), 41-49, April-June 2015
for critical systems since it is simple to implement and can be adopted for both hardware and software. The technique originally suggested by Von Neuman has drawn huge attention of researchers resulting in new improved methods for selecting correct and reliable output. The cost and complexity are the factors generally used to determine the number of redundant modules to be used for consensus. The utmost popular voting procedures for use are inexact majority and standard weighted average. Those widely in use these algorithms suffer inherent disadvantages. On other hand majority voting is found to be excellent in safety performance. The availability of the system would be compromised due to low probability of correct output while the other algorithm ensures better availability and poor safety performance. Many improvements have been suggested to overcome the limitations of the above mentioned voting techniques (LatifShabgahi and Hirst, 2005). Fuzzy logic aims at modelling the imprecise modes of reasoning emulating human approximation approach to make rational decision in an environment of uncertainty and imprecision. Fuzzy approach is an alternate design methodology enabling solutions to complex problems using probabilistic logic and probability theory. Latif-Shabgahi et al. has proposed a fuzzy voting scheme for fault tolerant system (2005). The proposed voter can be practicable substitute to the traditional inexact majority voting algorithm and the results have been studied and analysed using Fuzzy tools. In this work, we propose an improved fuzzy voter which can be employed by critical applications such as flight critical system which demands ultra-high safety performance and availability. The organization of the remaining paper is given as follows. Section 2 reviews related work in the area and discusses the fuzzy voter scheme in detail. A modified improved fuzzy voter is proposed in section 3 enumerating its advantages. The simulation results of the proposed scheme have been compared for safety performance and availability with respect to existing scheme suggested by Latif-Shabgahi (2005) in section 4. Finally section 5 summaries the outcome of the study.
2. REVIEW OF RELATED WORK One of the furthermost extensively used fault masking techniques in massive redundancy system is to deploy a voter to mitigate the effects of the fault. Since the output of the redundant modules are likely to differ within it. Therefore Researchers over the period have proposed numerous advanced techniques in order to improve on the drawback of the existing methods (Das and Bhattcharya, 2010; Manic and Frincke, 2001). Classical Majority voting uses results of redundant modules to arrive at correct output based on the threshold value of output generally implemented on m of n system (Parhami, 1992; Ross, 2008). Exact voting method falls short for dealing with noisy sensors output and rounding off errors due to use of floating point arithmetic. To deal with these limitations, many variants of inexact voting have been proposed in the literature to handle variety of situations to suite application specific voter characteristics (Krstic et al., 2005; Parhami, 1994). Weighted average voting method arrives at consensus by assigning weights to the modules depending on the closeness of their outputs. This method improves the availability however it has an adverse performance for modules with larger output differences (Parhami, 1994; Prakash and Vidyarthi, 2014). A standard inexact majority voting defines a limit on the deviation of module output with respect to the actual correct result. This is widely used for safety critical application. This technique ensures high safety performance with constrained on availability factor. An improved scheme where the participation score of modules is maintained over the operational period which gives the number of occasions the module contributed in the voting process. In an event of majority consensus, the output of module with highest score is selected from the group of majority consensus participating module. A voting approach using both weighted average and history based adaptive voting has also been reported to have a better availability as compared to the algorithms discussed (Ross, 2008). These voters generally use fixed thresholds values to arrive at the consensus between the redundant
Copyright © 2015, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited.
7 more pages are available in the full version of this document, which may be purchased using the "Add to Cart" button on the product's webpage: www.igi-global.com/article/an-improved-fuzzy-votingscheme-for-fault-tolerant-systems/131447?camid=4v1
This title is available in InfoSci-Journals, InfoSci-Journal Disciplines Computer Science, Security, and Information Technology. Recommend this product to your librarian: www.igi-global.com/e-resources/libraryrecommendation/?id=2
Related Content Reliability-Aware Proactive Energy Management in Hard Real-Time Systems: A Motivational Case Study Satyakiran Munaga and Francky Catthoor (2012). Technological Innovations in Adaptive and Dependable Systems: Advancing Models and Concepts (pp. 215-225).
www.igi-global.com/chapter/reliability-aware-proactive-energymanagement/63583?camid=4v1a Optimal Kernel and Wavelet Coefficients to Support Vector Regression Model and Wavelet Neural Network for Time Series Rainfall Prediction B. Kavitha Rani and A. Govardhan (2014). International Journal of Applied Evolutionary Computation (pp. 1-21).
www.igi-global.com/article/optimal-kernel-and-wavelet-coefficients-tosupport-vector-regression-model-and-wavelet-neural-network-for-timeseries-rainfall-prediction/126208?camid=4v1a Complexity, Information, and Robustness: The Role of Information “Barriers” in Boolean Networks Kurt A. Richardson (2008). Intelligent Complex Adaptive Systems (pp. 187-214).
www.igi-global.com/chapter/complexity-informationrobustness/24188?camid=4v1a
What Makes Reality: Ontological Classes and Rules Azamat Abdoullaev (2008). Reality, Universal Ontology and Knowledge Systems: Toward the Intelligent World (pp. 76-109).
www.igi-global.com/chapter/makes-reality-ontological-classesrules/28311?camid=4v1a
