Optimal Design of the κ -Out-of- n: G (F) Majority Voter

The κ -out-of- n : G (F) majority voter usually consists of n components (modules), and such a system is critical to ensure the correct operation of various computing systems for numerous critical applications. For a κ -out-of- n : G (F) majority voter, a specific number of the components are required to operate correctly for the overall system to function. To deal efficiently with the reliability evaluation of a general majority voter, a stochastic architecture can be adopted. The corresponding system reliability can be obtained through analyzing the output sequence. Usually, the system reliability is improved if more components or redundancies are used. Nevertheless, the consumed cost or required space also increases accordingly. In this paper, a tradeoff between the cost and reliability value was made to pursue the most desirable design. The relationship between the cost and corresponding component parameters is also discussed thoroughly in this paper. Then, to find the most cost-effective design, a new evaluation standard was proposed, referred to as the R-per-Cost. Furthermore, the optimal designs under different standards are presented for the investigated example. The results are also pursued with respect to an analysis of several case studies.