ANALYZING THE ERROR OF AN APPROXIMATE METHOD FOR COMPUTING PERFORMANCE-RELATED RELIABILITY

An approximate method for computing performance-related reliability of large-scale gracefully degrading systems was developed. Examples show that the error of this method decreases with the increase of the numb er of units composing the system, N. However, a theoretical analysis i s necessary to reveal the insight of this error reduction. In this pap er, the error of the approximate method is analyzed. The result shows that, in both cases, i.e. reconfiguration coverage C = 1 and C < 1, th e truncation error of the resulting performance-related reliability is of the order O(1N-2), showing a significant error decrease of the ne w method with the increase of N.