Customer Scheduling in Large Service Systems Under Model Uncertainty

Abstract

In the realm of many-server service systems, scheduling often necessitates the use of simplifying assumptions regarding service times to facilitate model development. However, empirical observations indicate that these assumptions may not accurately mirror real-world situations. In their paper titled “Customer Scheduling in Large Service Systems Under Model Uncertainty,” Chai, Sun, and Abouee-Mehrizi introduce an innovative approach to assist decision makers in devising high-quality scheduling policies for large service systems. This approach involves optimizing against an imaginary adversary through a robust control framework that is based on a manageable and simplified model. The imaginary adversary’s role is to exploit the potential vulnerabilities of a scheduling rule by dynamically perturbing the simplified model within an uncertainty set. This uncertainty set can be estimated using data-driven methods. Extensive numerical experiments, including a case study utilizing a data set from a U.S. call center, provide substantial evidence supporting the effectiveness of our framework.