People spend a long time waiting in line, e.g., at airports, banks, amusement parks, etc, and these long waits are often irritating and unpleasant. However, studies have shown that people’s frustration with waiting is reduced if they are given information, in advance, about their anticipated waits. One way to do so is via announcements of upcoming delays. To make effective announcements, we need to accurately predict future delays based on information about the present system state. We use queueing theory to propose such delay predictors, and show that our predictors are accurate and useful via both queueing analysis and simulation.
This research has both practical and academic dimensions. From a practical perspective, the alternative delay predictors that we propose are simple and effective. By simple, we mean that they rely on information which is easy to obtain in practice (queue length, previous waiting times, etc), and by effective we mean that they predict delays accurately. Thus, they can easily be used in a real-life context. From a theoretical perspective, we derived several queueing theoretic results, mostly regarding the accuracy of the different announcement schemes; such results provide a deeper understanding of the delay announcement problem in general.
Link to the publication’s UCL Discovery page