arXiv subjects

Apostolos Burnetas

Publications and source records attributed to Apostolos Burnetas.

6 recordsLinked to original sources

arXiv2026

Revenue and Social Welfare Maximization in Service Systems: Independent vs Grouped Customers

The revenue maximization problem in service systems with strategic customers is a central problem in the thread of Rational Queueing literature. The standard modeling assumption is that customers act independently and, therefore, the situation is modeled as a game among the customers and the service provider. In the present work, we focus on the case where the customers are grouped and controlled by their leader and the situation becomes a game between the customers' leader and the service provider. We examine the effect of customers' cooperation for the join-or-balk dilemma at the unobservable single-server Markovian queue and present the impact on the social welfare and service provider's revenue.

math.OC↗

arXiv2021

Single vs Dynamic Lead-Time Quotations in Make-To-Order Systems with Delay-Averse Customers

We develop a model for lead-time quotation in a Markovian make-to-order production or service system with strategic customers who exhibit risk aversion. Based on a CARA utility function of their net benefit, customers make individual decisions to join the system or balk by observing the state of the queue. The decisions of arriving customers result in a symmetric join/balk game. Regarding the firm's strategy, the provider announces a lead-time quotation for each state and a respective balking threshold. There is also a fixed entrance fee and compensation rate for the part of a customer' delay exceeding the quoted lead-time. Moreover, we consider the problem from the point of view of a social optimizer who maximizes the total net benefit of the system. We analyze the provider's and social optimizer's maximization problems and we consider two cases regarding the class of lead-time quotation policies, i.e., dynamic and single. We identify the optimal entrance thresholds in each case. Finally, through computational experiments we quantify the effect of risk aversion on the profits and the degree of flexibility that the compensation policy offers. It is shown that the detrimental effects of risk aversion can be addressed more efficiently for the provider's problem compared to the social optimizer's one. Furthermore, the profit loss when setting a single lead-time quote is generally small compared to the optimal dynamic quotation policy.

math.OC↗

arXiv2019

Lead-Time Quotations in Unobservable Make-To-Order Systems with Strategic Customers: Risk Aversion, Load Control and Profit Maximization

We develop a model for pricing, lead-time quotation and delay compensation in a Markovian make-to-order production or service system with strategic customers who exhibit risk aversion. Based on a concave utility function of their net benefit, customers make individual decisions to join the system or balk without observing the state of the queue. The decisions of arriving customers result in a symmetric join/balk game. Regarding the firm's strategy, the provider announces a fixed entrance fee, a lead-time quotation and a compensation rate for the part of a customer delay which exceeds the quoted lead-time. We analyze the effect of customer risk aversion and the compensation policy on the equilibrium join/balk strategies and the resulting input rates, and assess the flexibility of the provider in inducing a range of possible input rates under various constraints on the pricing/compensation policy. In numerical experiments we explore the behavior of pricing curves that reflect the provider's choices in inducing specific input rates. A key insight obtained from the analysis is that a main benefit of the lead-time and compensation option is to allow the entrance fee to remain high and the provider prefers strategies that lead to this direction.

math.OC↗

arXiv2018

Strategic Equilibria in Queues with Dynamic Service Rate and Full Information

We consider the problem of customer equilibrium behavior of a single server Markovian queue with dynamic control of the service rate. Customers arrive according a Poisson procedure and the system administrator makes a service rate choice between a low and a high value according to a $T$-threshold dynamic service policy, where the decision for switching to the higher service rate is made when the number of customers exceeds T without any additional cost. We assume that customers are identical and they are making join decisions regarding the maximization of their expected net benefit, receiving a fixed reward for service completion and incurring a waiting cost. In addition, we consider the observable case of the model where customers are fully informed on the service policy and the queue length upon arrival.

math.OC↗

arXiv2015

The Value of Service Rate Flexibility in an M/M/1 Queue with Admission Control

We consider a single server queueing system with admission control and the possibility to switch dynamically between a low and a high service rate, and examine the benefit of this service rate flexibility. We formulate a discounted Markov Decision Process model for the problem of joint admission and service control, and show that the optimal policy has a threshold structure for both controls. Regarding the benefit due to flexibility, we show that it is increasing in system congestion, and that its effect on the admission policy is to increase the admission threshold. We also derive a simple approximate condition between the admission reward and the relative cost of service rate increase, so that the service rate flexibility is beneficial. We finally show that the results extend to the expected average reward case.

math.OC↗

arXiv2012

Adaptive Policies for Sequential Sampling under Incomplete Information and a Cost Constraint

We consider the problem of sequential sampling from a finite number of independent statistical populations to maximize the expected infinite horizon average outcome per period, under a constraint that the expected average sampling cost does not exceed an upper bound. The outcome distributions are not known. We construct a class of consistent adaptive policies, under which the average outcome converges with probability 1 to the true value under complete information for all distributions with finite means. We also compare the rate of convergence for various policies in this class using simulation.

stat.ML↗