Smoothing voter: a novel voting algorithm for handling multiple errors in fault-tolerant control systems

Voting algorithms are used to arbitrate between variant results in a wide range of highly dependable real-time control applications. These applications include N-Modular Redundant hardware systems and diversely designed software systems based on N-Version Programming. The most sophisticated and complex voting algorithms can even tolerate malicious (or Byzantine) variant errors. Voting algorithms can be implemented in either hardware or software depending on the characteristics of the application and the type of voter selected. The behaviour of voting algorithms in multiple error scenarios is considered in this article. Complete disagreement is defined as those cases where no two variant results are the same. A novel algorithm for real-time control applications, the smoothing voter, is introduced and its behaviour compared with three previously published voters. Software implemented error-injection tests, reported here, show that the smoothing voter achieves a compromise between the result selection capabilities of the median voter and the safety features of the majority voter. The smoothing voter is an appropriate voter for applications in which maximising the number of correct outputs and minimising the number of benign errors of the system is the main concern, and a slight degradation in safety can be tolerated. q 2003 Published by Elsevier Science B.V.