Automatic Fuzzy Parameter Selection in Dynamic Fuzzy Voter for Safety Critical Systems

Automatic Fuzzy Parameter Selection in Dynamic Fuzzy Voter for Safety Critical Systems

PhaniKumar Singamsetty, SeethaRamaiah Panchumarthy
Copyright: © 2012 |Pages: 23
DOI: 10.4018/ijfsa.2012040104
(Individual Articles)
No Current Special Offers


The main objective of this research paper is designing automatic fuzzy parameter selection based dynamic fuzzy voter for safety critical systems with limited system knowledge. Existing fuzzy voters for controlling safety critical systems and sensor fusion are surveyed and safety performance is empirically 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. In this paper, a dynamic or automatic fuzzy parameter selection method for fuzzy voters is proposed based on the statistical parameters of the local set of data in each voting cycle. Safety performance is empirically evaluated by running the static and dynamic fuzzy voters on a simulated triple modular redundant (TMR) system for 10000 voting cycles. Experimental results show that proposed Dynamic fuzzy voter is giving almost 100% safety if two of the three modules of the TMR System are error free. Dynamic voter is designed in such a way that it can be plugged in and used in any safety critical system without having any knowledge regarding the data produced and their ranges.
Article Preview


Safety critical systems are the systems which may lead to hazards, loss of lives or great damage to the property if they fail. There are different domains in which safety critical control systems are used: (automotives) drive-by-wire systems, brake by wire systems used in cars; (medicine) infusion pumps, cancer radiation therapy machines, etc.; (military and space applications) rocket launchers, satellite launchers, etc.; and (industrial process control) robotics and consumer electronic appliances. There is a need to increase the reliability, availability and safety in all these applications. Faults that occur in these applications may lead to hazardous situations. If a single module or channel is used and when it becomes faulty due to some noise the system may fail and hazard may occur. Hence N – modular redundancy or N-version programming along with voting technique is used to mask the faults in the faulty environments as given in Laprie (1985) and Johnson (1989).

There are different architectural patterns as mentioned in Kumar et al. (2011), in which redundant modules with a voter are used in the safety-critical systems. All the N-modules or N-versions as shown in Chen and Avizienis (1978) are designed by different teams to meet the same specifications. All these modules take the same input data, process it and generate the results which will be passed to the voter. The voter has to mask the fault by isolating or avoiding the faulty module and the correct value has to be picked by the voter.

There are different types of voting algorithms mentioned in the literature Latif-Shabgahi et al. (2004). Some voting algorithms like majority, plurality voters as shown in Blough and Sullivan (1990) generate the output if the majority or required numbers of inputs to the voter are matched; otherwise it will generate no output so that the system can be taken to the fail safe state. Adaptive majority voting algorithm designed by Latif-Shabgahi and Bennett (1999) gives better performance by using history records. But for some safety-critical systems, there may not be any fail safe state. In such systems, the voter has to generate some value as the output using some methods like amalgamating the outputs or results of all modules, which is called as result amalgamation as shown in Lorczak et al. (1989). Median, average, weighted average voters as given in Latif-Shabgahi (2004) are some examples for the voters which amalgamate the inputs of the voter and generate some value as the voter output. History based weighted average voters as given in Latif-Shabgahi et al. (2001) consider the history of the modules and for the highly reliable module greater weight is given. History based weighted voter designed by Singamsetty and Panchumarthy (2011) considers the history and module agreeability product to calculate the weights of the modules which are used in weighted average calculation of the voter output. In Das and Battacharya (2010), modified history based weighted average voting with soft dynamic threshold is designed. In this work, the threshold is calculated based upon the notional correct output of the voter. It is difficult to predict the voter output before only to decide the threshold. It is a major limitation in this voter. In Zarafshan et al. (2010), a neural network based voter is designed and the neural network is trained using feed forward error back propagation algorithm. It is time taking process to train the network.

The main objective of this research paper is designing automatic fuzzy parameter selection based self configurable dynamic fuzzy voter for safety critical systems.

This rest of the paper is organized as follows. Existing fuzzy voters are surveyed and their performance is evaluated and limitations are identified. An automatic fuzzy parameter selection method for fuzzy voters is then proposed. Design variations of static and dynamic fuzzy voters are reviewed. Test harness used in the experimentation is given and experimental results are discussed. Performance of the dynamic fuzzy voter is evaluated and compared with the existing static fuzzy voter. Based on the analysis results, some points are made. Finally, the concluding remarks and some directions for future research are given.

Complete Article List

Search this Journal:
Volume 13: 1 Issue (2024)
Volume 12: 1 Issue (2023)
Volume 11: 4 Issues (2022)
Volume 10: 4 Issues (2021)
Volume 9: 4 Issues (2020)
Volume 8: 4 Issues (2019)
Volume 7: 4 Issues (2018)
Volume 6: 4 Issues (2017)
Volume 5: 4 Issues (2016)
Volume 4: 4 Issues (2015)
Volume 3: 4 Issues (2013)
Volume 2: 4 Issues (2012)
Volume 1: 4 Issues (2011)
View Complete Journal Contents Listing