Asymptotically Optimal Clearing Control of Backlogs in Multiclass Processing Systems
研究多类排队系统如何最优清理大量积压订单，对两类客户证明折扣cμ/θ规则的静态优先级策略渐近最优，对多类客户提出最佳努力规则并验证其有效性。
Processing systems, such as make-to-order production or service systems, are often faced with backlogged demand, which results in a prolonged period in which the system is congested, although it has sufficient processing capacity to handle all newly arriving demand. In “Asymptotically Optimal Clearing Control of Backlogs in Multiclass Processing Systems,” Yu, Iravani, and Perry consider a processing system, modeled by a multiclass queueing model, that faces the problem of optimally clearing a large backlog from several classes of customers (or orders). For the special case of two classes, the authors prove that a static priority policy following a discounted cμ/θ rule is asymptotically optimal. When there are more than two classes of customers, the authors show that any admissible control that follows the best-effort rule becomes asymptotically optimal after a relatively short time. An extensive numerical study shows that these proposed policies are effective and provides guidance on when to choose among the policies in practice.