Adversarial issues in reliability

Abstract

Many reliability problems involve two or more agents with conflicting interests whose decisions affect the performance of the system at hand. Examples of such problems relevant in management practice abound and include acceptance sampling, life testing, software testing, optimal maintenance, reliability demonstration, warranties and insurance. Most earlier attempts in such problems have focused on game theoretic approaches based on Nash equilibria and related concepts. However, these require strong common knowledge assumptions which do not frequently hold in practice. We provide an alternative framework based on adversarial risk analysis to deal with such problems which avoids the strong common knowledge assumptions of game theory. We illustrate the framework through acceptance sampling and life testing problems.

Advances in Bayesian decision making in reliability

“…Besides its focus on Bayesian decision making, this article differs from the other reviews in reliability due to its consideration of adversarial issues discussed in Section 4 . As noted by Rios Insua et al (2018) , the ARA approach discussed here can be extended to other areas such as warranty policies and software testing which include more than two adversaries. For example, Ruggeri and Soyer (2018) point out that the optimal release problem in software engineering, which has been considered from a game theoretic viewpoint by Zeephongsekul and Chiera (1995) , can be formulated using the ARA framework.…”

“…Reliability problems involving a single decision maker can be formulated and solved with the Bayesian decision-theoretic framework. However, as noted by Rios Insua, Ruggeri, Soyer, and Rasines (2018) , there are reliability problems that involve two or more decision makers, possibly with opposed interests. One of the earliest Bayesian approaches in adversarial reliability settings is due to Lindley and Singpurwalla (1991) , Lindley and Singpurwalla (1993) who considered problems in acceptance sampling and life testing.…”

“…provide details on assessment of the probability distributions of the consumer and the utility function U C ( c , θ ). The ARA framework presented above has been used by RiosInsua et al (2018) in Bernoulli acceptance sampling and exponential life testing.…”